_id stringlengths 4 10 | text stringlengths 0 18.4k | title stringlengths 0 8.56k |
|---|---|---|
d257079142 | Humans excel at lifelong learning, as the brain has evolved to be robust to distribution shifts and noise in our ever-changing environment. Deep neural networks (DNNs), however, exhibit catastrophic forgetting and the learned representations drift drastically as they encounter a new task. This alludes to a different errorbased learning mechanism in the brain. Unlike DNNs, where learning scales linearly with the magnitude of the error, the sensitivity to errors in the brain decreases as a function of their magnitude. To this end, we propose ESMER which employs a principled mechanism to modulate error sensitivity in a dual-memory rehearsalbased system. Concretely, it maintains a memory of past errors and uses it to modify the learning dynamics so that the model learns more from small consistent errors compared to large sudden errors. We also propose Error-Sensitive Reservoir Sampling to maintain episodic memory, which leverages the error history to pre-select low-loss samples as candidates for the buffer, which are better suited for retaining information. Empirical results show that ESMER effectively reduces forgetting and abrupt drift in representations at the task boundary by gradually adapting to the new task while consolidating knowledge. Remarkably, it also enables the model to learn under high levels of label noise, which is ubiquitous in real-world data streams. Code: https://github.com/NeurAI-Lab/ESMER * Contributed equally. | ERROR SENSITIVITY MODULATION BASED EXPERI- ENCE REPLAY: MITIGATING ABRUPT REPRESENTA- TION DRIFT IN CONTINUAL LEARNING |
d1204679 | The linear layer is one of the most pervasive modules in deep learning representations. However, it requires O(N 2 ) parameters and O(N 2 ) operations. These costs can be prohibitive in mobile applications or prevent scaling in many domains. Here, we introduce a deep, differentiable, fully-connected neural network module composed of diagonal matrices of parameters, A and D, and the discrete cosine transform C. The core module, structured as ACDC −1 , has O(N ) parameters and incurs O(N log N ) operations. We present theoretical results showing how deep cascades of ACDC layers approximate linear layers. ACDC is, however, a stand-alone module and can be used in combination with any other types of module. In our experiments, we show that it can indeed be successfully interleaved with ReLU modules in convolutional neural networks for image recognition. Our experiments also study critical factors in the training of these structured modules, including initialization and depth. Finally, this paper also points out avenues for implementing the complex version of ACDC using photonic devices. | ACDC: A STRUCTURED EFFICIENT LINEAR LAYER |
d5096141 | Hash codes are a very efficient data representation needed to be able to cope with the ever growing amounts of data. We introduce a random forest semantic hashing scheme with information-theoretic code aggregation, showing for the first time how random forest, a technique that together with deep learning have shown spectacular results in classification, can also be extended to large-scale retrieval. Traditional random forest fails to enforce the consistency of hashes generated from each tree for the same class data, i.e., to preserve the underlying similarity, and it also lacks a principled way for code aggregation across trees. We start with a simple hashing scheme, where independently trained random trees in a forest are acting as hashing functions. We the propose a subspace model as the splitting function, and show that it enforces the hash consistency in a tree for data from the same class. We also introduce an information-theoretic approach for aggregating codes of individual trees into a single hash code, producing a near-optimal unique hash for each class. Experiments on large-scale public datasets are presented, showing that the proposed approach significantly outperforms state-of-the-art hashing methods for retrieval tasks. | Under review as a workshop contribution at ICLR 2015 RANDOM FORESTS CAN HASH |
d225075982 | Empirical studies demonstrate that the performance of neural networks improves with increasing number of parameters. In most of these studies, the number of parameters is increased by increasing the network width. This begs the question: Is the observed improvement due to the larger number of parameters, or is it due to the larger width itself? We compare different ways of increasing model width while keeping the number of parameters constant. We show that for models initialized with a random, static sparsity pattern in the weight tensors, network width is the determining factor for good performance, while the number of weights is secondary, as long as the model achieves high training accuarcy. As a step towards understanding this effect, we analyze these models in the framework of Gaussian Process kernels. We find that the distance between the sparse finite-width model kernel and the infinite-width kernel at initialization is indicative of model performance. | Published as a conference paper at ICLR 2021 ARE WIDER NETS BETTER GIVEN THE SAME NUMBER OF PARAMETERS? |
d257255014 | Recent works on neural contextual bandits have achieved compelling performances due to their ability to leverage the strong representation power of neural networks (NNs) for reward prediction. Many applications of contextual bandits involve multiple agents who collaborate without sharing raw observations, thus giving rise to the setting of federated contextual bandits. Existing works on federated contextual bandits rely on linear or kernelized bandits, which may fall short when modeling complex real-world reward functions. So, this paper introduces the federated neural-upper confidence bound (FN-UCB) algorithm. To better exploit the federated setting, FN-UCB adopts a weighted combination of two UCBs: UCB a allows every agent to additionally use the observations from the other agents to accelerate exploration (without sharing raw observations), while UCB b uses an NN with aggregated parameters for reward prediction in a similar way to federated averaging for supervised learning. Notably, the weight between the two UCBs required by our theoretical analysis is amenable to an interesting interpretation, which emphasizes UCB a initially for accelerated exploration and relies more on UCB b later after enough observations have been collected to train the NNs for accurate reward prediction (i.e., reliable exploitation). We prove sub-linear upper bounds on both the cumulative regret and the number of communication rounds of FN-UCB, and empirically demonstrate its competitive performance. . FL-NTK: A neural tangent kernel-based framework for federated learning analysis. In | Published as a conference paper at ICLR 2023 FEDERATED NEURAL BANDITS |
d68222714 | Deep Neural Networks (DNNs) excel on many complex perceptual tasks but it has proven notoriously difficult to understand how they reach their decisions. We here introduce a high-performance DNN architecture on ImageNet whose decisions are considerably easier to explain. Our model, a simple variant of the ResNet-50 architecture called BagNet, classifies an image based on the occurrences of small local image features without taking into account their spatial ordering. This strategy is closely related to the bag-of-feature (BoF) models popular before the onset of deep learning and reaches a surprisingly high accuracy on ImageNet (87.6% top-5 for 33 × 33 px features and Alexnet performance for 17 × 17 px features). The constraint on local features makes it straight-forward to analyse how exactly each part of the image influences the classification. Furthermore, the BagNets behave similar to state-of-the art deep neural networks such as VGG-16, ResNet-152 or DenseNet-169 in terms of feature sensitivity, error distribution and interactions between image parts. This suggests that the improvements of DNNs over previous bag-of-feature classifiers in the last few years is mostly achieved by better fine-tuning rather than by qualitatively different decision strategies. | APPROXIMATING CNNS WITH BAG-OF-LOCAL- FEATURES MODELS WORKS SURPRISINGLY WELL ON IMAGENET |
d253801755 | We consider the general problem of recovering a high-dimensional signal from noisy quantized measurements. Quantization, especially coarse quantization such as 1-bit sign measurements, leads to severe information loss and thus a good prior knowledge of the unknown signal is helpful for accurate recovery. Motivated by the power of score-based generative models (SGM, also known as diffusion models) in capturing the rich structure of natural signals beyond simple sparsity, we propose an unsupervised data-driven approach called quantized compressed sensing with SGM (QCS-SGM), where the prior distribution is modeled by a pre-trained SGM. To perform posterior sampling, an annealed pseudo-likelihood score called noise perturbed pseudo-likelihood score is introduced and combined with the prior score of SGM. The proposed QCS-SGM applies to an arbitrary number of quantization bits. Experiments on a variety of baseline datasets demonstrate that the proposed QCS-SGM significantly outperforms existing state-of-the-art algorithms by a large margin for both in-distribution and out-of-distribution samples. Moreover, as a posterior sampling method, QCS-SGM can be easily used to obtain confidence intervals or uncertainty estimates of the reconstructed results. The code is available at https://github.com/mengxiangming/QCS-SGM. Figure 1: Reconstructed images of our QCS-SGM for one FFHQ 256 × 256 high-resolution RGB test image (N = 256 × 256 × 3 = 196608 pixels) from noisy heavily quantized (1bit, 2-bit and 3-bit) CS 8× measurements y = Q(Ax + n), i.e., M = 24576 N . The measurement matrix A ∈ R M ×N is i.i.d. Gaussian, i.e., Aij ∼ N (0, 1 M ), and a Gaussian noise n is added with standard deviation σ = 10 −3 . compressed sensing under asymmetric noise. In Conference on Learning Theory, pp. 152-192. PMLR, 2016. , et al. Optimization with sparsity-inducing penalties. Foundations and Trends® in Machine Learning, 4(1):1-106, 2012. Amir Beck and Marc Teboulle. A fast iterative shrinkage-thresholding algorithm for linear inverse problems. SIAM journal on imaging sciences, 2(1):183-202, 2009.Ashish Bora, Ajil Jalal, Eric Price, and Alexandros G Dimakis. Compressed sensing using generative . Accurate sampling using langevin dynamics. Physical Review E, 75(5):056707, 2007. Emmanuel J Candès and Michael B Wakin. An introduction to compressive sampling. IEEE signal processing magazine, 25(2):21-30, 2008. Emmanuel J Candès, Justin Romberg, and Terence Tao. Robust uncertainty principles: Exact signal reconstruction from highly incomplete frequency information. IEEE Transactions on information theory, 52(2):489-509, 2006. Hyungjin Chung, Byeongsu Sim, Dohoon Ryu, and Jong Chul Ye. Improving diffusion models for inverse problems using manifold constraints. arXiv preprint arXiv:2206.00941, 2022. Wei Dai and Olgica Milenkovic. Information theoretical and algorithmic approaches to quantized compressive sensing. IEEE transactions on communications, 59(7):1857-1866, 2011. Giannis Daras, Yuval Dagan, Alex Dimakis, and Constantinos Daskalakis. Score-guided intermediate level optimization: Fast langevin mixing for inverse problems. | Published as a conference paper at ICLR 2023 QUANTIZED COMPRESSED SENSING WITH SCORE- BASED GENERATIVE MODELS |
d3338083 | The ability to synthesize realistic patterns of neural activity is crucial for studying neural information processing. Here we used the Generative Adversarial Networks (GANs) framework to simulate the concerted activity of a population of neurons. We adapted the Wasserstein-GAN variant to facilitate the generation of unconstrained neural population activity patterns while still benefiting from parameter sharing in the temporal domain. We demonstrate that our proposed GAN, which we termed Spike-GAN, generates spike trains that match accurately the first-and second-order statistics of datasets of tens of neurons and also approximates well their higher-order statistics. We applied Spike-GAN to a real dataset recorded from salamander retina and showed that it performs as well as state-ofthe-art approaches based on the maximum entropy and the dichotomized Gaussian frameworks. Importantly, Spike-GAN does not require to specify a priori the statistics to be matched by the model, and so constitutes a more flexible method than these alternative approaches. Finally, we show how to exploit a trained Spike-GAN to construct 'importance maps' to detect the most relevant statistical structures present in a spike train. Spike-GAN provides a powerful, easy-to-use technique for generating realistic spiking neural activity and for describing the most relevant features of the large-scale neural population recordings studied in modern systems neuroscience. | Published as a conference paper at ICLR 2018 SYNTHESIZING REALISTIC NEURAL POPULATION ACTIVITY PATTERNS USING GENERATIVE ADVERSARIAL NETWORKS |
d252531892 | In cooperative multi-agent reinforcement learning (MARL), combining value decomposition with actor-critic enables agents to learn stochastic policies, which are more suitable for the partially observable environment. Given the goal of learning local policies that enable decentralized execution, agents are commonly assumed to be independent of each other, even in centralized training. However, such an assumption may prohibit agents from learning the optimal joint policy. To address this problem, we explicitly take the dependency among agents into centralized training. Although this leads to the optimal joint policy, it may not be factorized for decentralized execution. Nevertheless, we theoretically show that from such a joint policy, we can always derive another joint policy that achieves the same optimality but can be factorized for decentralized execution. To this end, we propose multi-agent conditional policy factorization (MACPF), which takes more centralized training but still enables decentralized execution. We empirically verify MACPF in various cooperative MARL tasks and demonstrate that MACPF achieves better performance or faster convergence than baselines. Our code is available at https | MORE CENTRALIZED TRAINING, STILL DECENTRAL- IZED EXECUTION: MULTI-AGENT CONDITIONAL POL- ICY FACTORIZATION |
d238582653 | Pretraining language models with next-token prediction on massive text corpora has delivered phenomenal zero-shot, few-shot, transfer learning and multi-tasking capabilities on both generative and discriminative language tasks. Motivated by this success, we explore a Vector-quantized Image Modeling (VIM) approach that involves pretraining a Transformer to predict rasterized image tokens autoregressively. The discrete image tokens are encoded from a learned Vision-Transformerbased VQGAN (ViT-VQGAN). We first propose multiple improvements over vanilla VQGAN from architecture to codebook learning, yielding better efficiency and reconstruction fidelity. The improved ViT-VQGAN further improves vectorquantized image modeling tasks, including unconditional, class-conditioned image generation and unsupervised representation learning. When trained on Im-ageNet at 256 × 256 resolution, we achieve Inception Score (IS) of 175.1 and Fréchet Inception Distance (FID) of 4.17, a dramatic improvement over the vanilla VQGAN, which obtains 70.6 and 17.04 for IS and FID, respectively. Based on ViT-VQGAN and unsupervised pretraining, we further evaluate the pretrained Transformer by averaging intermediate features, similar to Image GPT (iGPT). This ImageNet-pretrained VIM-L significantly beats iGPT-L on linear-probe accuracy from 60.3% to 73.2% for a similar model size. VIM-L also outperforms iGPT-XL which is trained with extra web image data and larger model size.arXiv:2110.04627v3 [cs.CV] 5 Jun 2022Published as a conference paper at ICLR 2022 Figure 1: Overview of ViT-VQGAN (left) and Vector-quantized Image Modeling (right) for both image generation and image understanding.Remarkable image generation results have been achieved by pre-quantizing images into discrete latent variables and modeling them autoregressively, including VQVAE (Oord et al., 2017), DALL-E (Ramesh et al., 2021) and VQGAN (Esser et al., 2021). In these approaches, a convolution neural network (CNN) is learned to auto-encode an image and a second stage CNN or Transformer is learned to model the density of encoded latent variables. These have been proved effective for image generation, but few studies have evaluated the learned representation in discriminative tasks (Ramesh et al., 2021; Esser et al., 2021).We explore an approach we refer to as Vector-quantized Image Modeling (VIM) and apply it to both image generation and image understanding tasks. VIM follows a two-stage approach: | VECTOR-QUANTIZED IMAGE MODELING WITH IM- PROVED VQGAN |
d46929526 | In recent years, deep reinforcement learning has been shown to be adept at solving sequential decision processes with high-dimensional state spaces such as in the Atari games. Many reinforcement learning problems, however, involve highdimensional discrete action spaces as well as high-dimensional state spaces. This paper considers entropy bonus, which is used to encourage exploration in policy gradient. In the case of high-dimensional action spaces, calculating the entropy and its gradient requires enumerating all the actions in the action space and running forward and backpropagation for each action, which may be computationally infeasible. We develop several novel unbiased estimators for the entropy bonus and its gradient. We apply these estimators to several models for the parameterized policies, including Independent Sampling, CommNet, Autoregressive with Modified MDP, and Autoregressive with LSTM. Finally, we test our algorithms on two environments: a multi-hunter multi-rabbit grid game and a multi-agent multi-arm bandit problem. The results show that our entropy estimators substantially improve performance with marginal additional computational cost.Preprint. Work in progress. | Efficient Entropy for Policy Gradient with Multidimensional Action Space |
d252846362 | The graph Transformer emerges as a new architecture and has shown superior performance on various graph mining tasks. In this work, we observe that existing graph Transformers treat nodes as independent tokens and construct a single long sequence composed of all node tokens so as to train the Transformer model, causing it hard to scale to large graphs due to the quadratic complexity on the number of nodes for the self-attention computation. To this end, we propose a Neighborhood Aggregation Graph Transformer (NAGphormer) that treats each node as a sequence containing a series of tokens constructed by our proposed Hop2Token module. For each node, Hop2Token aggregates the neighborhood features from different hops into different representations and thereby produces a sequence of token vectors as one input. In this way, NAGphormer could be trained in a mini-batch manner and thus could scale to large graphs. Moreover, we mathematically show that as compared to a category of advanced Graph Neural Networks (GNNs), the decoupled Graph Convolutional Network, NAGphormer could learn more informative node representations from the multi-hop neighborhoods. Extensive experiments on benchmark datasets from small to large are conducted to demonstrate that NAGphormer consistently outperforms existing graph Transformers and mainstream GNNs. | NAGPHORMER: A TOKENIZED GRAPH TRANS- FORMER FOR NODE CLASSIFICATION IN LARGE GRAPHS |
d211082896 | We introduce a new routing algorithm for capsule networks, in which a child capsule is routed to a parent based only on agreement between the parent's state and the child's vote. The new mechanism 1) designs routing via inverted dot-product attention; 2) imposes Layer Normalization as normalization; and 3) replaces sequential iterative routing with concurrent iterative routing. When compared to previously proposed routing algorithms, our method improves performance on benchmark datasets such as CIFAR-10 and CIFAR-100, and it performs at-par with a powerful CNN (ResNet-18) with 4× fewer parameters. On a different task of recognizing digits from overlayed digit images, the proposed capsule model performs favorably against CNNs given the same number of layers and neurons per layer. We believe that our work raises the possibility of applying capsule networks to complex real-world tasks. Our code is publicly available at: https://github. com/apple/ml-capsules-inverted-attention-routing. | Published as a conference paper at ICLR 2020 CAPSULES WITH INVERTED DOT-PRODUCT ATTENTION ROUTING |
d247476275 | Common image-to-image translation methods rely on joint training over data from both source and target domains. The training process requires concurrent access to both datasets, which hinders data separation and privacy protection; and existing models cannot be easily adapted for translation of new domain pairs. We present Dual Diffusion Implicit Bridges (DDIBs), an image translation method based on diffusion models, that circumvents training on domain pairs. Image translation with DDIBs relies on two diffusion models trained independently on each domain, and is a two-step process: DDIBs first obtain latent encodings for source images with the source diffusion model, and then decode such encodings using the target model to construct target images. Both steps are defined via ordinary differential equations (ODEs), thus the process is cycle consistent only up to discretization errors of the ODE solvers. Theoretically, we interpret DDIBs as concatenation of source to latent, and latent to target Schrödinger Bridges, a form of entropy-regularized optimal transport, to explain the efficacy of the method. Experimentally, we apply DDIBs on synthetic and high-resolution image datasets, to demonstrate their utility in a wide variety of translation tasks and their inherent optimal transport properties. | Published as a conference paper at ICLR 2023 DUAL DIFFUSION IMPLICIT BRIDGES FOR IMAGE-TO-IMAGE TRANSLATION |
d5668935 | Multilabel image annotation is one of the most important challenges in computer vision with many real-world applications. While existing work usually use conventional visual features for multilabel annotation, features based on Deep Neural Networks have shown potential to significantly boost performance. In this work, we propose to leverage the advantage of such features and analyze key components that lead to better performances. Specifically, we show that a significant performance gain could be obtained by combining convolutional architectures with approximate top-k ranking objectives, as thye naturally fit the multilabel tagging problem. Our experiments on the NUS-WIDE dataset outperforms the conventional visual features by about 10%, obtaining the best reported performance in the literature. | Deep Convolutional Ranking for Multilabel Image Annotation |
d247595391 | We introduce the concept of provably robust adversarial examples for deep neural networks -connected input regions constructed from standard adversarial examples which are guaranteed to be robust to a set of real-world perturbations (such as changes in pixel intensity and geometric transformations). We present a novel method called PARADE for generating these regions in a scalable manner which works by iteratively refining the region initially obtained via sampling until a refined region is certified to be adversarial with existing state-of-the-art verifiers. At each step, a novel optimization procedure is applied to maximize the region's volume under the constraint that the convex relaxation of the network behavior with respect to the region implies a chosen bound on the certification objective. Our experimental evaluation shows the effectiveness of PARADE: it successfully finds large provably robust regions including ones containing ≈ 10 573 adversarial examples for pixel intensity and ≈ 10 599 for geometric perturbations. The provability enables our robust examples to be significantly more effective against state-of-theart defenses based on randomized smoothing than the individual attacks used to construct the regions.Published as a conference paper at ICLR 2022 cation techniques making it easily extendable to new adversarial attack models. We make the code of PARADE available at https://github.com/eth-sri/parade.git • A thorough evaluation of PARADE, demonstrating it can generate provable regions containing ≈ 10 573 concrete adversarial points for pixel intensity changes, in ≈ 2 minutes, and ≈ 10 599 concrete points for geometric transformations, in ≈ 20 minutes, on a challenging CIFAR10 network. We also demonstrate that our robust adversarial examples are significantly more effective against state-of-the-art defenses based on randomized smoothing than the individual attacks used to construct the regions.BACKGROUNDWe now discuss the background necessary for the remainder of the paper. We consider a neural network f : R n0 → R n l with l layers, n 0 input neurons and n l output classes. While our method can handle arbitrary activations, we focus on networks with the widely-used ReLU activation. The network classifies an input x to class y(x) with the largest corresponding output value, i.e.,Note for brevity we omit the argument to y when it is clear from the context.NEURAL NETWORK CERTIFICATIONIn this work, we rely on existing state-of-the-art neural network certification methods based on convex relaxations to prove that the adversarial examples produced by our algorithm are robust. These certification methods take a convex input region I ⊂ R n0 and prove that every point in I is classified as the target label y t by f . They propagate the set I through the layers of the network, producing a convex region that covers all possible values of the output neurons . Robustness follows by proving that, for all combinations of output neuron values in this region, the output neuron corresponding to class y t has a larger value than the one corresponding to any other class y = y t .Commonly, one proves this property by computing a function L y : R n0 → R for each label y = y t , such that, for all x ∈ I, we have L y (x) ≤ [f (x)] yt − [f (x)] y . For each L y , one computes min x∈I L y (x) to obtain a global lower bound that is true for all x ∈ I. If we obtain positive bounds for all y = y t , robustness is proven. To simplify notation, we will say that the certification objective L(x) is the function L y (x) with the smallest minimum value on I. We will call its corresponding minimum value the certification error. We require L y (x) to be a linear function of x. This requirement is consistent with many popular certification algorithms based on convex relaxation, such as CROWN (Zhang et al., 2018), DeepZ (Singh et al., 2018a), and DeepPoly (Singh et al., 2019). Without loss of generality, for the rest of this paper, we will treat DeepPoly as our preferred certification method.CERTIFICATION AGAINST GEOMETRIC TRANSFORMATIONSDeepPoly operates over specifications based on linear constraints over input pixels for verification. These constraints are straightforward to provide for simple pixel intensity transformations such as adversarial patches(Chiang et al., 2020)and L ∞ (Carlini & Wagner, 2017) perturbations that provide a closed-form formula for the input region. However, geometric transformations do not yield such linear regions. To prove the robustness of our generated examples to geometric transformations, we rely on DeepG (Balunović et al., 2019) which, given a range of geometric transformation parameters, creates an overapproximation of the set of input images generated by the geometric perturbations. DeepG then leverages DeepPoly to certify the input image region. When generating our geometric robust examples, we work directly in the geometric parameter space and, thus, our input region I and the inputs to our certification objective L(x) are also in geometric space. Despite this change, as our approach is agnostic to the choice of the verifier, in the remainder of the paper we will assume the certification is done using DeepPoly and not DeepG, unless otherwise stated.RANDOMIZED SMOOTHINGRandomized smoothing(Lécuyer et al., 2019;Cohen et al., 2019) is a provable defense mechanism against adversarial attacks. For a chosen standard deviation σ and neural network f as defined above, randomized smoothing computes a smoothed classifier g based on f , such that g(x) = argmax c P(y(x + ) = c) with random Gaussian noise ∼ N (0, σ 2 I). This construction of g allows | Published as a conference paper at ICLR 2022 PROVABLY ROBUST ADVERSARIAL EXAMPLES |
d249461729 | For unsupervised pretraining, mask-reconstruction pretraining (MRP) approaches, e.g. MAE (He et al., 2021) and data2vec (Baevski et al., 2022), randomly mask input patches and then reconstruct the pixels or semantic features of these masked patches via an auto-encoder. Then for a downstream task, supervised fine-tuning the pretrained encoder remarkably surpasses the conventional "supervised learning" (SL) trained from scratch. However, it is still unclear 1) how MRP performs semantic feature learning in the pretraining phase and 2) why it helps in downstream tasks.To solve these problems, we first theoretically show that on an auto-encoder of a two/one-layered convolution encoder/decoder, MRP can capture all discriminative features of each potential semantic class in the pretraining dataset. Then considering the fact that the pretraining dataset is of huge size and high diversity and thus covers most features in downstream dataset, in fine-tuning phase, the pretrained encoder can capture as much features as it can in downstream datasets, and would not lost these features with theoretical guarantees. In contrast, SL only randomly captures some features due to lottery ticket hypothesis. So MRP provably achieves better performance than SL on the classification tasks. Experimental results testify to our data assumptions and also our theoretical implications. * Equal contribution. Pan Jiachun did this work during an internship at Sea AI Lab. | Published as a conference paper at ICLR 2023 TOWARDS UNDERSTANDING WHY MASK RECON- STRUCTION PRETRAINING HELPS IN DOWNSTREAM TASKS |
d239016586 | Multi-head, key-value attention is the backbone of the widely successful Transformer model and its variants. This attention mechanism uses multiple parallel key-value attention blocks (called heads), each performing two fundamental computations: (1) search -selection of a relevant entity from a set via query-key interactions, and (2) retrieval -extraction of relevant features from the selected entity via a value matrix. Importantly, standard attention heads learn a rigid mapping between search and retrieval. In this work, we first highlight how this static nature of the pairing can potentially: (a) lead to learning of redundant parameters in certain tasks, and (b) hinder generalization. To alleviate this problem, we propose a novel attention mechanism, called Compositional Attention, that replaces the standard head structure. The proposed mechanism disentangles search and retrieval and composes them in a dynamic, flexible and context-dependent manner through an additional soft competition stage between the query-key combination and value pairing. Through a series of numerical experiments, we show that it outperforms standard multi-head attention on a variety of tasks, including some out-of-distribution settings. Through our qualitative analysis, we demonstrate that Compositional Attention leads to dynamic specialization based on the type of retrieval needed. Our proposed mechanism generalizes multi-head attention, allows independent scaling of search and retrieval, and can easily be implemented in lieu of standard attention heads in any network architecture. | Published as a conference paper at ICLR 2022 COMPOSITIONAL ATTENTION: DISENTANGLING SEARCH AND RETRIEVAL |
d235078790 | Overparameterized Neural Networks (NN) display state-of-the-art performance. However, there is a growing need for smaller, energy-efficient, neural networks to be able to use machine learning applications on devices with limited computational resources. A popular approach consists of using pruning techniques. While these techniques have traditionally focused on pruning pre-trained NN (LeCun et al., 1990;Hassibi et al., 1993), recent work by Lee et al. (2018) has shown promising results when pruning at initialization. However, for Deep NNs, such procedures remain unsatisfactory as the resulting pruned networks can be difficult to train and, for instance, they do not prevent one layer from being fully pruned. In this paper, we provide a comprehensive theoretical analysis of Magnitude and Gradient based pruning at initialization and training of sparse architectures. This allows us to propose novel principled approaches which we validate experimentally on a variety of NN architectures. | Published as a conference paper at ICLR 2021 ROBUST PRUNING AT INITIALIZATION |
d233033714 | We find that the way we choose to represent data labels can have a profound effect on the quality of trained models. For example, training an image classifier to regress audio labels rather than traditional categorical probabilities produces a more reliable classification. This result is surprising, considering that audio labels are more complex than simpler numerical probabilities or text. We hypothesize that high dimensional, high entropy label representations are generally more useful because they provide a stronger error signal. We support this hypothesis with evidence from various label representations including constant matrices, spectrograms, shuffled spectrograms, Gaussian mixtures, and uniform random matrices of various dimensionalities. Our experiments reveal that high dimensional, high entropy labels achieve comparable accuracy to text (categorical) labels on the standard image classification task, but features learned through our label representations exhibit more robustness under various adversarial attacks and better effectiveness with a limited amount of training data. These results suggest that label representation may play a more important role than previously thought. | Published as a conference paper at ICLR 2021 BEYOND CATEGORICAL LABEL REPRESENTATIONS FOR IMAGE CLASSIFICATION |
d258352540 | The success of the Adam optimizer on a wide array of architectures has made it the default in settings where stochastic gradient descent (SGD) performs poorly. However, our theoretical understanding of this discrepancy is lagging, preventing the development of significant improvements on either algorithm. Recent work advances the hypothesis that Adam and other heuristics like gradient clipping outperform SGD on language tasks because the distribution of the error induced by sampling has heavy tails. This suggests that Adam outperform SGD because it uses a more robust gradient estimate. We evaluate this hypothesis by varying the batch size, up to the entire dataset, to control for stochasticity. We present evidence that stochasticity and heavy-tailed noise are not major factors in the performance gap between SGD and Adam. Rather, Adam performs better as the batch size increases, while SGD is less effective at taking advantage of the reduction in noise. This raises the question as to why Adam outperforms SGD in the full-batch setting. Through further investigation of simpler variants of SGD, we find that the behavior of Adam with large batches is similar to sign descent with momentum. | Published as a conference paper at ICLR 2023 NOISE IS NOT THE MAIN FACTOR BEHIND THE GAP BETWEEN SGD AND ADAM ON TRANSFORMERS, BUT SIGN DESCENT MIGHT BE |
d252568019 | In reinforcement learning for safety-critical settings, it is often desirable for the agent to obey safety constraints at all points in time, including during training. We present a novel neurosymbolic approach called SPICE to solve this safe exploration problem. SPICE uses an online shielding layer based on symbolic weakest preconditions to achieve a more precise safety analysis than existing tools without unduly impacting the training process. We evaluate the approach on a suite of continuous control benchmarks and show that it can achieve comparable performance to existing safe learning techniques while incurring fewer safety violations. Additionally, we present theoretical results showing that SPICE converges to the optimal safe policy under reasonable assumptions. * equal advising 1 SPICE is available at https | Published as a conference paper at ICLR 2023 GUIDING SAFE EXPLORATION WITH WEAKEST PRECONDITIONS |
d249089172 | When a dynamical system can be modeled as a sequence of observations, Granger causality is a powerful approach for detecting predictive interactions between its variables. However, traditional Granger causal inference has limited utility in domains where the dynamics need to be represented as directed acyclic graphs (DAGs) rather than as a linear sequence, such as with cell differentiation trajectories. Here, we present GrID-Net, a framework based on graph neural networks with lagged message passing for Granger causal inference on DAG-structured systems. Our motivating application is the analysis of single-cell multimodal data to identify genomic loci that mediate the regulation of specific genes. To our knowledge, GrID-Net is the first single-cell analysis tool that accounts for the temporal lag between a genomic locus becoming accessible and its downstream effect on a target gene's expression. We applied GrID-Net on multimodal single-cell assays that profile chromatin accessibility (ATAC-seq) and gene expression (RNA-seq) in the same cell and show that it dramatically outperforms existing methods for inferring regulatory locus-gene links, achieving up to 71% greater agreement with independent population genetics-based estimates. By extending Granger causality to DAG-structured dynamical systems, our work unlocks new domains for causal analyses and, more specifically, opens a path towards elucidating gene regulatory interactions relevant to cellular differentiation and complex human diseases at unprecedented scale and resolution. 1 | Published as a conference paper at ICLR 2022 GRANGER CAUSAL INFERENCE ON DAGS IDENTIFIES GENOMIC LOCI REGULATING TRANSCRIPTION |
d13687188 | Teaching plays a very important role in our society, by spreading human knowledge and educating our next generations. A good teacher will select appropriate teaching materials, impact suitable methodologies, and set up targeted examinations, according to the learning behaviors of the students. In the field of artificial intelligence, however, one has not fully explored the role of teaching, and pays most attention to machine learning. In this paper, we argue that equal attention, if not more, should be paid to teaching, and furthermore, an optimization framework (instead of heuristics) should be used to obtain good teaching strategies. We call this approach "learning to teach". In the approach, two intelligent agents interact with each other: a student model (which corresponds to the learner in traditional machine learning algorithms), and a teacher model (which determines the appropriate data, loss function, and hypothesis space to facilitate the training of the student model). The teacher model leverages the feedback from the student model to optimize its own teaching strategies by means of reinforcement learning, so as to achieve teacher-student co-evolution. To demonstrate the practical value of our proposed approach, we take the training of deep neural networks (DNN) as an example, and show that by using the learning to teach techniques, we are able to use much less training data and fewer iterations to achieve almost the same accuracy for different kinds of DNN models (e.g., multi-layer perceptron, convolutional neural networks and recurrent neural networks) under various machine learning tasks (e.g., image classification and text understanding). | Published as a conference paper at ICLR 2018 LEARNING TO TEACH |
d3279351 | E-commerce companies such as Amazon, Alibaba and Flipkart process billions of orders every year. However, these orders represent only a small fraction of all plausible orders. Exploring the space of all plausible orders could help us better understand the relationships between the various entities in an e-commerce ecosystem, namely the customers and the products they purchase. In this paper, we propose a Generative Adversarial Network (GAN) for orders made in e-commerce websites. Once trained, the generator in the GAN could generate any number of plausible orders. Our contributions include: (a) creating a dense and low-dimensional representation of e-commerce orders, (b) train an ecommerceGAN (ecGAN) with real orders to show the feasibility of the proposed paradigm, and (c) train an ecommerce-conditional-GAN (ec 2 GAN) to generate the plausible orders involving a particular product. We propose several qualitative methods to evaluate ecGAN and demonstrate its effectiveness. The ec 2 GAN is used for various kinds of characterization of possible orders involving a product that has just been introduced into the e-commerce system. The proposed approach ec 2 GAN performs significantly better than the baseline in most of the scenarios. | eCommerceGAN : A Generative Adversarial Network for E-commerce |
d258298544 | A common assumption when training embodied agents is that the impact of taking an action is stable; for instance, executing the "move ahead" action will always move the agent forward by a fixed distance, perhaps with some small amount of actuator-induced noise. This assumption is limiting; an agent may encounter settings that dramatically alter the impact of actions: a move ahead action on a wet floor may send the agent twice as far as it expects and using the same action with a broken wheel might transform the expected translation into a rotation. Instead of relying that the impact of an action stably reflects its pre-defined semantic meaning, we propose to model the impact of actions on-the-fly using latent embeddings. By combining these latent action embeddings with a novel, transformerbased, policy head, we design an Action Adaptive Policy (AAP). We evaluate our AAP on two challenging visual navigation tasks in the AI2-THOR and Habitat environments and show that our AAP is highly performant even when faced, at inference-time with missing actions and, previously unseen, perturbed action space. Moreover, we observe significant improvement in robustness against these actions when evaluating in real-world scenarios. On evaluation of embodied navigation agents. arXiv, 2018. 8, 16Somrita Banerjee, James Harrison, P Michael Furlong, and Marco Pavone. Adaptive meta-learning for identification of rover-terrain dynamics. arXiv, 2020. 3 | Published as a conference paper at ICLR 2023 MOVING FORWARD BY MOVING BACKWARD: EMBED- DING ACTION IMPACT OVER ACTION SEMANTICS/projects/action-adaptive-policy |
d232307616 | Recognizing relations between entities is a pivotal task of relational learning. Learning relation representations from distantly-labeled datasets is difficult because of the abundant label noise and complicated expressions in human language. This paper aims to learn predictive, interpretable, and robust relation representations from distantly-labeled data that are effective in different settings, including supervised, distantly supervised, and few-shot learning. Instead of solely relying on the supervision from noisy labels, we propose to learn prototypes for each relation from contextual information to best explore the intrinsic semantics of relations. Prototypes are representations in the feature space abstracting the essential semantics of relations between entities in sentences. We learn prototypes based on objectives with clear geometric interpretation, where the prototypes are unit vectors uniformly dispersed in a unit ball, and statement embeddings are centered at the end of their corresponding prototype vectors on the surface of the ball. This approach allows us to learn meaningful, interpretable prototypes for the final classification. Results on several relation learning tasks show that our model significantly outperforms the previous state-of-the-art models. We further demonstrate the robustness of the encoder and the interpretability of prototypes with extensive experiments. * Equal contribution. † Corresponding author. | Published as a conference paper at ICLR 2021 PROTOTYPICAL REPRESENTATION LEARNING FOR RE- LATION EXTRACTION |
d53389725 | Efforts to reduce the numerical precision of computations in deep learning training have yielded systems that aggressively quantize weights and activations, yet employ wide high-precision accumulators for partial sums in inner-product operations to preserve the quality of convergence. The absence of any framework to analyze the precision requirements of partial sum accumulations results in conservative design choices. This imposes an upper-bound on the reduction of complexity of multiply-accumulate units. We present a statistical approach to analyze the impact of reduced accumulation precision on deep learning training. Observing that a bad choice for accumulation precision results in loss of information that manifests itself as a reduction in variance in an ensemble of partial sums, we derive a set of equations that relate this variance to the length of accumulation and the minimum number of bits needed for accumulation. We apply our analysis to three benchmark networks: CIFAR-10 ResNet 32, ImageNet ResNet 18 and Ima-geNet AlexNet. In each case, with accumulation precision set in accordance with our proposed equations, the networks successfully converge to the single precision floating-point baseline. We also show that reducing accumulation precision further degrades the quality of the trained network, proving that our equations produce tight bounds. Overall this analysis enables precise tailoring of computation hardware to the application, yielding area-and power-optimal systems. | ACCUMULATION BIT-WIDTH SCALING FOR ULTRA- LOW PRECISION TRAINING OF DEEP NETWORKS |
d244116787 | a) class: band aid, spurious feature: fingers, -41.54% (b) class: space bar, spurious feature: keys, -46.15% (c) class: plate, spurious feature: food, -32.31% (d) class: butterfly, spurious feature: flowers, -21.54% (e) class: potter's wheel, spurious feature: vase, -21.54% | Published as a conference paper at ICLR 2022 SALIENT IMAGENET: HOW TO DISCOVER SPURIOUS FEATURES IN DEEP LEARNING? |
d238634783 | Spatial convolutions 1 are widely used in numerous deep video models. It fundamentally assumes spatio-temporal invariance, i.e., using shared weights for every location in different frames. This work presents Temporally-Adaptive Convolutions (TAdaConv) for video understanding 2 , which shows that adaptive weight calibration along the temporal dimension is an efficient way to facilitate modelling complex temporal dynamics in videos. Specifically, TAdaConv empowers the spatial convolutions with temporal modelling abilities by calibrating the convolution weights for each frame according to its local and global temporal context. Compared to previous temporal modelling operations, TAdaConv is more efficient as it operates over the convolution kernels instead of the features, whose dimension is an order of magnitude smaller than the spatial resolutions. Further, the kernel calibration brings an increased model capacity. We construct TAda2D and TAda-ConvNeXt networks by replacing the 2D convolutions in ResNet and ConvNeXt with TAdaConv, which leads to at least on par or better performance compared to state-of-the-art approaches on multiple video action recognition and localization benchmarks. We also demonstrate that as a readily plug-in operation with negligible computation overhead, TAdaConv can effectively improve many existing video models with a convincing margin. | Published as a conference paper at ICLR 2022 TADA! TEMPORALLY-ADAPTIVE CONVOLUTIONS FOR VIDEO UNDERSTANDING |
d252683760 | Monotonic linear interpolation (MLI) -on the line connecting a random initialization with the minimizer it converges to, the loss and accuracy are monotonic -is a phenomenon that is commonly observed in the training of neural networks. Such a phenomenon may seem to suggest that optimization of neural networks is easy. In this paper, we show that the MLI property is not necessarily related to the hardness of optimization problems, and empirical observations on MLI for deep neural networks depend heavily on the biases. In particular, we show that interpolating both weights and biases linearly leads to very different influences on the final output, and when different classes have different last-layer biases on a deep network, there will be a long plateau in both the loss and accuracy interpolation (which existing theory of MLI cannot explain). We also show how the last-layer biases for different classes can be different even on a perfectly balanced dataset using a simple model. Empirically we demonstrate that similar intuitions hold on practical networks and realistic datasets. | Published as a conference paper at ICLR 2023 PLATEAU IN MONOTONIC LINEAR INTERPOLATION - A "BIASED" VIEW OF LOSS LANDSCAPE FOR DEEP NETWORKS |
d257913759 | We improve upon previous oblivious sketching and turnstile streaming results for 1 and logistic regression, giving a much smaller sketching dimension achieving O(1)-approximation and yielding an efficient optimization problem in the sketch space. Namely, we achieve for any constant c > 0 a sketching dimension ofÕ(d 1+c ) for 1 regression andÕ(µd 1+c ) for logistic regression, where µ is a standard measure that captures the complexity of compressing the data. For 1-regression our sketching dimension is near-linear and improves previous work which either required Ω(log d)-approximation with this sketching dimension, or required a larger poly(d) number of rows. Similarly, for logistic regression previous work had worse poly(µd) factors in its sketching dimension. We also give a tradeoff that yields a 1 + ε approximation in input sparsity time by increasing the total size to (d log(n)/ε) O(1/ε) for 1 and to (µd log(n)/ε) O(1/ε) for logistic regression. Finally, we show that our sketch can be extended to approximate a regularized version of logistic regression where the data-dependent regularizer corresponds to the variance of the individual logistic losses.IntroductionWe consider logistic regression in distributed and streaming environments. A key tool for solving these problems is a distribution over random oblivious linear maps S ∈ R r×n which have the property that, for a given n × d matrix X, where we assume the labels for the rows of X have been multiplied into X, given only SX one can efficiently and approximately solve the logistic regression problem. The fact that S does not depend on X is what is referred to as S being oblivious, which is important in distributed and streaming tasks since one can choose S without first needing to read the input data. The fact that S is a linear map is also important for such tasks, since given SX (1) and SX (2) , one can add these to obtain S(X (1) + X (2) ), which allows for positive or negative updates to entries of the input in a stream, or across multiple servers in the arbitrary partition model of communication, see, e.g., [Woodruff, 2014] for a discussion of data stream and communication models.An important goal is to minimize the sketching dimension r of the sketching matrix S, as this translates into the memory required of a streaming algorithm and the communication cost of a distributed algorithm. At the same time, one would like the approximation factor that one obtains via this approach to be as small as possible. Specifically we develop and improve oblivious sketching for the most important robust linear regression variant, namely 1 regression, and for logistic regression, which is a generalized linear model of high importance for binary classification and estimation of Bernoulli probabilities. Sketching supports very fast updates which is desirable for performing robust and generalized regression in high-velocity data processing applications, for instance in physical experiments and other resource constraint settings, cf. [Munteanu et al., 2021, Munteanu, 2023.We focus on the case where the number n of data points is very large, i.e., n d. In this case, applying a standard algorithm directly is not a viable option since it is either too slow or even becomes impossible when * Dortmund | Almost Linear Constant-Factor Sketching for 1 and Logistic Regression |
d26238954 | We formulate language modeling as a matrix factorization problem, and show that the expressiveness of Softmax-based models (including the majority of neural language models) is limited by a Softmax bottleneck. Given that natural language is highly context-dependent, this further implies that in practice Softmax with distributed word embeddings does not have enough capacity to model natural language. We propose a simple and effective method to address this issue, and improve the state-of-the-art perplexities on Penn Treebank and WikiText-2 to 47.69 and 40.68 respectively. 1 * Equal contribution. Ordering determined by dice rolling. 1 Code is available at httpsUnder review as a conference paper at ICLR 2018 matrices that have much larger normalized singular values and thus much higher rank than Softmax and other baselines on real-world datasets.We evaluate our proposed approach on standard language modeling benchmarks. MoS substantially improves over the current state-of-the-art results on benchmarks, by up to 3.6 points in terms of perplexity, reaching perplexities 47.69 on Penn Treebank and 40.68 on WikiText-2. We further apply MoS to a dialog dataset and show improved performance over Softmax and other baselines.Our contribution is two-fold. First, we identify the Softmax bottleneck by formulating language modeling as a matrix factorization problem. Second, we propose a simple and effective method that substantially improves over the current state-of-the-art results. | Under review as a conference paper at ICLR 2018 BREAKING THE SOFTMAX BOTTLENECK: A HIGH-RANK RNN LANGUAGE MODEL |
d4009713 | Over the last years, deep convolutional neural networks (ConvNets) have transformed the field of computer vision thanks to their unparalleled capacity to learn high level semantic image features. However, in order to successfully learn those features, they usually require massive amounts of manually labeled data, which is both expensive and impractical to scale. Therefore, unsupervised semantic feature learning, i.e., learning without requiring manual annotation effort, is of crucial importance in order to successfully harvest the vast amount of visual data that are available today. In our work we propose to learn image features by training Con-vNets to recognize the 2d rotation that is applied to the image that it gets as input. We demonstrate both qualitatively and quantitatively that this apparently simple task actually provides a very powerful supervisory signal for semantic feature learning. We exhaustively evaluate our method in various unsupervised feature learning benchmarks and we exhibit in all of them state-of-the-art performance. Specifically, our results on those benchmarks demonstrate dramatic improvements w.r.t. prior state-of-the-art approaches in unsupervised representation learning and thus significantly close the gap with supervised feature learning. For instance, in PASCAL VOC 2007 detection task our unsupervised pre-trained AlexNet model achieves the state-of-the-art (among unsupervised methods) mAP of 54.4% that is only 2.4 points lower from the supervised case. We get similarly striking results when we transfer our unsupervised learned features on various other tasks, such as ImageNet classification, PASCAL classification, PASCAL segmentation, and CIFAR-10 classification. The code and models of our paper will be published on: https://github.com/gidariss/FeatureLearningRotNet. | Published as a conference paper at ICLR 2018 UNSUPERVISED REPRESENTATION LEARNING BY PRE- DICTING IMAGE ROTATIONS |
d232240244 | Data augmentation is an effective technique to improve the generalization of deep neural networks. However, previous data augmentation methods usually treat the augmented samples equally without considering their individual impacts on the model. To address this, for the augmented samples from the same training example, we propose to assign different weights to them. We construct the maximal expected loss which is the supremum over any reweighted loss on augmented samples. Inspired by adversarial training, we minimize this maximal expected loss (MMEL) and obtain a simple and interpretable closed-form solution: more attention should be paid to augmented samples with large loss values (i.e., harder examples). Minimizing this maximal expected loss enables the model to perform well under any reweighting strategy. The proposed method can generally be applied on top of any data augmentation methods. Experiments are conducted on both natural language understanding tasks with token-level data augmentation, and image classification tasks with commonly-used image augmentation techniques like random crop and horizontal flip. Empirical results show that the proposed method improves the generalization performance of the model. * This work is done when Mingyang Yi is an intern at Huawei Noah's Ark Lab. | Published as a conference paper at ICLR 2021 REWEIGHTING AUGMENTED SAMPLES BY MINIMIZ- ING THE MAXIMAL EXPECTED LOSS |
d2524977 | In this report, we describe a Theano-based AlexNet (Krizhevsky et al., 2012) implementation and its naive data parallelism on multiple GPUs. Our performance on 2 GPUs is comparable with the state-of-art Caffe library (Jia et al., 2014) run on 1 GPU. To the best of our knowledge, this is the first open-source Python-based AlexNet implementation to-date. | THEANO-BASED LARGE-SCALE VISUAL RECOGNI- TION WITH MULTIPLE GPUS |
d257232949 | Recently, both Contrastive Learning (CL) and Mask Image Modeling (MIM) demonstrate that self-supervision is powerful to learn good representations. However, naively combining them is far from success. In this paper, we start by making the empirical observation that a naive joint optimization of CL and MIM losses leads to conflicting gradient directions -more severe as the layers go deeper. This motivates us to shift the paradigm from combining loss at the end, to choosing the proper learning method per network layer. Inspired by experimental observations, we find that MIM and CL are suitable to lower and higher layers, respectively. We hence propose to combine them in a surprisingly simple, "sequential cascade" fashion: early layers are first trained under one MIM loss, on top of which latter layers continue to be trained under another CL loss. The proposed Layer Grafted Pre-training learns good visual representations that demonstrate superior label efficiency in downstream applications, in particular yielding strong few-shot performance besides linear evaluation. For instance, on ImageNet-1k, Layer Grafted Pre-training yields 65.5% Top-1 accuracy in terms of 1% few-shot learning with ViT-B/16, which improves MIM and CL baselines by 14.4% and 2.1% with no bells and whistles. The code is available at https://github. com/VITA-Group/layerGraftedPretraining_ICLR23.git. * Part of this work was conducted during a summer internship at Microsoft.Published as a conference paper at ICLR 2023 conflicts in the gradient directions, as the layers go deeper (seeFigure 1). That causes considerable hurdles for the (pre-)training to effectively proceed.We are then inspired to ask: if the two losses conflict when both are placed at the end, how about placing them differently, such as appending them to different layers? Based on experimental observations, it appears that lower layers tend to learn better from the MIM loss in order to capture local spatial details; while higher layers tend to benefit more from the CL loss in order to learn semantically-aware grouping and invariance. Inspired by so, we propose a simple MIM→CL Grafting idea to combine the bests of both worlds: (step i) first training the lower layers with MIM loss and fixing their weights, on top of which (step ii) higher layer weights continue to be trained under another CL loss. This simple cascaded training idea neatly separates MIM and CL losses to avoid their conflicts against each other if placed together; each loss is also strategically placed to pre-training its most suitable portion. Practically, we "'smooth out" the grafting by allowing lower layers to be slowly tuned in step ii. Our ablation experiments also find that the order of grafting matters, i.e., reversing MIM/CL loss locations and performing CL→MIM will considerably damage the performance. The contributions of this paper are summarized as follows:• We propose Layer Grafted Pre-training, a principled framework to merge MIM and CL, improving representation learning beyond both, with no bells and whistles.• We investigate the different preferences of lower and higher layers towards CL and MIM losses, and show the order of grafting to matter. | LAYER GRAFTED PRE-TRAINING: BRIDGING CON- TRASTIVE LEARNING AND MASKED IMAGE MODEL- ING FOR LABEL-EFFICIENT REPRESENTATIONS |
d252815657 | Transformers flexibly operate over sets of real-valued vectors representing taskspecific entities and their attributes, where each vector might encode one wordpiece token and its position in a sequence, or some piece of information that carries no position at all. But as set processors, standard transformers are at a disadvantage in reasoning over more general graph-structured data where nodes represent entities and edges represent relations between entities. To address this shortcoming, we generalize transformer attention to consider and update edge vectors in each transformer layer. We evaluate this relational transformer on a diverse array of graph-structured tasks, including the large and challenging CLRS Algorithmic Reasoning Benchmark. There, it dramatically outperforms state-of-theart graph neural networks expressly designed to reason over graph-structured data. Our analysis demonstrates that these gains are attributable to relational attention's inherent ability to leverage the greater expressivity of graphs over sets. | Published as a conference paper at ICLR 2023 RELATIONAL ATTENTION: GENERALIZING TRANSFORMERS FOR GRAPH-STRUCTURED TASKS |
d3507990 | In few-shot classification, we are interested in learning algorithms that train a classifier from only a handful of labeled examples. Recent progress in few-shot classification has featured meta-learning, in which a parameterized model for a learning algorithm is defined and trained on episodes representing different classification problems, each with a small labeled training set and its corresponding test set. In this work, we advance this few-shot classification paradigm towards a scenario where unlabeled examples are also available within each episode. We consider two situations: one where all unlabeled examples are assumed to belong to the same set of classes as the labeled examples of the episode, as well as the more challenging situation where examples from other distractor classes are also provided. To address this paradigm, we propose novel extensions of Prototypical Networks (Snell et al., 2017) that are augmented with the ability to use unlabeled examples when producing prototypes. These models are trained in an end-to-end way on episodes, to learn to leverage the unlabeled examples successfully. We evaluate these methods on versions of the Omniglot and miniImageNet benchmarks, adapted to this new framework augmented with unlabeled examples. We also propose a new split of ImageNet, consisting of a large set of classes, with a hierarchical structure. Our experiments confirm that our Prototypical Networks can learn to improve their predictions due to unlabeled examples, much like a semi-supervised algorithm would. * Equal contribution. 1 See the following blog post for an overview: | Published as a conference paper at ICLR 2018 META-LEARNING FOR SEMI-SUPERVISED FEW-SHOT CLASSIFICATION |
d3790787 | Structured prediction energy networks (SPENs; Belanger & McCallum 2016) use neural network architectures to define energy functions that can capture arbitrary dependencies among parts of structured outputs. Prior work used gradient descent for inference, relaxing the structured output to a set of continuous variables and then optimizing the energy with respect to them. We replace this use of gradient descent with a neural network trained to approximate structured argmax inference. This "inference network" outputs continuous values that we treat as the output structure. We develop large-margin training criteria for joint training of the structured energy function and inference network. On multi-label classification we report speed-ups of 10-60x compared to(Belanger et al., 2017)while also improving accuracy. For sequence labeling with simple structured energies, our approach performs comparably to exact inference while being much faster at test time. We then demonstrate improved accuracy by augmenting the energy with a "label language model" that scores entire output label sequences, showing it can improve handling of long-distance dependencies in part-of-speech tagging. Finally, we show how inference networks can replace dynamic programming for test-time inference in conditional random fields, suggestive for their general use for fast inference in structured settings. | Published as a conference paper at ICLR 2018 LEARNING APPROXIMATE INFERENCE NETWORKS FOR STRUCTURED PREDICTION |
d6458072 | We propose several simple approaches to training deep neural networks on data with noisy labels. We introduce an extra noise layer into the network which adapts the network outputs to match the noisy label distribution. The parameters of this noise layer can be estimated as part of the training process and involve simple modifications to current training infrastructures for deep networks. We demonstrate the approaches on several datasets, including large scale experiments on the ImageNet classification benchmark, showing how additional noisy data can improve state-of-the-art recognition models.The effect of label noise is well studied in common classifiers (e.g., SVMs, kNN, logistic regression), and their label noise robust variants have been proposed. See[5]for comprehensive review. A more recent work [2] proposed a generic unbiased estimator for binary classification with noisy labels. They employ a surrogate cost function that can be expressed by a weighted sum of the original cost functions, and gave theoretical bounds on the performance. In this paper, we will also consider this idea and extend it multiclass.A cost function similar to ours is proposed in [2] to make logistic regression robust to label noise. They also proposed a learning algorithm for noise parameters. However, we consider deep networks, a more powerful and complex classifier than logistic regression, and propose a different learning algorithm for noise parameters that is more suited for back-propagation training.Considering the recent success of deep learning[8,17,16], there are very few works about deep learning from noisy labels. In[11,9], noise modeling is incorporated to neural network in the same way as our proposed model. However, only binary classification is considered in[11], and [9] assumed symmetric label noise (noise is independent of the true label). Therefore, there is only a single noise parameter, which can be tuned by cross-validation. In this paper, we consider multiclass classification and assume more realistic asymmetric label noise, which makes it impossible to use cross-validation to adjust noise parameters (there can be a million parameters).ApproachIn this paper, we consider two approaches to make an existing classification model, which we call the base model, robust against noisy labels: bottom-up and top-down noise models. In the bottomup model, we add an additional layer to the model that changes the label probabilities output by the base model so it would better match to noisy labels. Top-down model, on other hand, changes given noisy labels before feeding them to the base model. Both models require a noise model for training, so we will give an easy way to estimate noise levels using clean data. Also, it is possible to learn noise distribution from noisy data in the bottom-up model. Although only deep neural networks are used in our experiments, the both approaches can be applied to any classification model with a cross entropy cost.Bottom-up Noise ModelWe assume that label noise is random conditioned on the true class, but independent of the input x (see[10]for more detail about this type of noise). Based on this assumption, we add an additional layer to a deep network (seeFigure 1) that changes its output so it would better match to the noisy labels. The weights of this layer corresponds to the probabilities of a certain class being mislabeled to another class. Because those probabilities are often unknown, we will show how estimate them from additional clean data, or from the noisy data itself.Let D be the true data distribution generating correctly labeled samples (x, y * ), where x is an input vector and y * is the corresponding label. However, we only observe noisy labeled samples (x,ỹ) that generated from a some noisy distributionD. We assume that the label noise is random conditioned on the true labels. Then, the noise distribution can be parameterized by a matrix Q = {q ji }: q ji := p(ỹ = j|y * = i). Q is a probability matrix because its elements are positive and each column sums to one. The probability of input x being labeled as j inD is given bywhere p(y * = i|x, θ) is the probabilistic output of the base model with parameters θ. If the true noise distribution is known, we can modify this for noisy labeled data. During training, Q will act as an adapter that transforms the model's output to better match the noisy labels.AbstractThe abstract goes here.AbstractThe abstract goes here. | Learning from Noisy Labels with Deep Neural Networks |
d249152128 | Auxiliary objectives, supplementary learning signals that are introduced to help aid learning on data-starved or highly complex end-tasks, are commonplace in machine learning. Whilst much work has been done to formulate useful auxiliary objectives, their construction is still an art which proceeds by slow and tedious handdesign. Intuition for how and when these objectives improve end-task performance has also had limited theoretical backing. In this work, we present an approach for automatically generating a suite of auxiliary objectives. We achieve this by deconstructing existing objectives within a novel unified taxonomy, identifying connections between them, and generating new ones based on the uncovered structure. Next, we theoretically formalize widely-held intuitions about how auxiliary learning improves generalization on the end-task. This leads us to a principled and efficient algorithm for searching the space of generated objectives to find those most useful to a specified end-task. With natural language processing (NLP) as our domain of study, we demonstrate that our automated auxiliary learning pipeline leads to strong improvements over competitive baselines across continued training experiments on a pre-trained model on 5 NLP tasks 1 . | Published as a conference paper at ICLR 2023 AANG: AUTOMATING AUXILIARY LEARNING |
d246861899 | Point cloud analysis is challenging due to irregularity and unordered data structure. To capture the 3D geometries, prior works mainly rely on exploring sophisticated local geometric extractors using convolution, graph, or attention mechanisms. These methods, however, incur unfavorable latency during inference, and the performance saturates over the past few years. In this paper, we present a novel perspective on this task. We notice that detailed local geometrical information probably is not the key to point cloud analysis -we introduce a pure residual MLP network, called PointMLP, which integrates no "sophisticated" local geometrical extractors but still performs very competitively. Equipped with a proposed lightweight geometric affine module, PointMLP delivers the new stateof-the-art on multiple datasets. On the real-world ScanObjectNN dataset, our method even surpasses the prior best method by 3.3% accuracy. We emphasize that PointMLP achieves this strong performance without any sophisticated operations, hence leading to a superior inference speed. Compared to most recent CurveNet, PointMLP trains 2× faster, tests 7× faster, and is more accurate on ModelNet40 benchmark. We hope our PointMLP may help the community towards a better understanding of point cloud analysis. The code is available at | Published as a conference paper at ICLR 2022 RETHINKING NETWORK DESIGN AND LOCAL GEOM- ETRY IN POINT CLOUD: A SIMPLE RESIDUAL MLP FRAMEWORK |
d235376887 | Recently, numerous machine learning based methods for combinatorial optimization problems have been proposed that learn to construct solutions in a sequential decision process via reinforcement learning. While these methods can be easily combined with search strategies like sampling and beam search, it is not straightforward to integrate them into a high-level search procedure offering strong search guidance.Bello et al. (2016)propose active search, which adjusts the weights of a (trained) model with respect to a single instance at test time using reinforcement learning. While active search is simple to implement, it is not competitive with state-of-the-art methods because adjusting all model weights for each test instance is very time and memory intensive. Instead of updating all model weights, we propose and evaluate three efficient active search strategies that only update a subset of parameters during the search. The proposed methods offer a simple way to significantly improve the search performance of a given model and outperform state-of-the-art machine learning based methods on combinatorial problems, even surpassing the well-known heuristic solver LKH3 on the capacitated vehicle routing problem. Finally, we show that (efficient) active search enables learned models to effectively solve instances that are much larger than those seen during training. | Published as a conference paper at ICLR 2022 EFFICIENT ACTIVE SEARCH FOR COMBINATORIAL OPTIMIZATION PROBLEMS |
d1882658 | In this paper we consider a problem of searching a space of predictive models for a given training data set. We propose an iterative procedure for deriving a sequence of improving models and a corresponding sequence of sets of non-linear features on the original input space. After a finite number of iterations N , the non-linear features become 2 N -degree polynomials on the original space. We show that in a limit of an infinite number of iterations derived non-linear features must form an associative algebra: a product of two features is equal to a linear combination of features from the same feature space for any given input point. Because each iteration consists of solving a series of convex problems that contain all previous solutions, the likelihood of the models in the sequence is increasing with each iteration while the dimension of the model parameter space is set to a limited controlled value. | |
d9369284 | Workshop track -ICLR 2017 CHANGING MODEL BEHAVIOR AT TEST-TIME USING REINFORCEMENT LEARNING | |
d3630111 | We consider the setting of an agent with a fixed body interacting with an unknown and uncertain external world. We show that models trained to predict proprioceptive information about the agent's body come to represent objects in the external world. In spite of being trained with only internally available signals, these dynamic body models come to represent external objects through the necessity of predicting their effects on the agent's own body. That is, the model learns holistic persistent representations of objects in the world, even though the only training signals are body signals. Our dynamics model is able to successfully predict distributions over 132 sensor readings over 100 steps into the future and we demonstrate that even when the body is no longer in contact with an object, the latent variables of the dynamics model continue to represent its shape. We show that active data collection by maximizing the entropy of predictions about the bodytouch sensors, proprioception and vestibular information-leads to learning of dynamic models that show superior performance when used for control. We also collect data from a real robotic hand and show that the same models can be used to answer questions about properties of objects in the real world. Videos with qualitative results of our models are available at https://goo.gl/mZuqAV. | Published as a conference paper at ICLR 2018 LEARNING AWARENESS MODELS |
d247362882 | Product quantization (PQ) coupled with a space rotation, is widely used in modern approximate nearest neighbor (ANN) search systems to significantly compress the disk storage for embeddings and speed up the inner product computation. Existing rotation learning methods, however, minimize quantization distortion for fixed embeddings, which are not applicable to an end-to-end training scenario where embeddings are updated constantly. In this paper, based on geometric intuitions from Lie group theory, in particular the special orthogonal group SO(n), we propose a family of block Givens coordinate descent algorithms to learn rotation matrix that are provably convergent on any convex objectives. Compared to the state-of-the-art SVD method, the Givens algorithms are much more parallelizable, reducing runtime by orders of magnitude on modern GPUs, and converge more stably according to experimental studies. They further improve upon vanilla product quantization significantly in an end-to-end training scenario.SinceJegou et al. (2010)first introduced PQ and Asymmetric Distance Computation (ADC) from signal processing to ANN search problem, there have been multiple lines of research focused on improving PQ based indexes. We briefly summarize the main directions below.Coarse Quantization, also referred to as inverted file (IVF), is introduced in the original work ofJegou et al. (2010)to first learn a full vector quantization (referred to as coarse quantization) by k-means clustering and then perform product quantization over the residual of the coarse quantization. This enables non-exhaustive search of only a subsets of the clusters, which allows ANN to retrieve billions of embeddings in tens of milliseconds. More work has been done in this direction, including Inverted Multi-Index (IMI) (Lempitsky, 2012); Implementation Optimization efforts have mainly been spent on the computation of ADC, including using Hamming distance for fast pruning(Douze et al., 2016), an efficient GPU implementation of ADC lookup(Johnson et al., 2019), and SIMD-based computation for lower bounds of ADC(André et al., 2015). Our proposed method is fully compatible with all these implementation optimizations, since we are solely focused on the rotation matrix learning algorithms. | Published as a conference paper at ICLR 2022 GIVENS COORDINATE DESCENT METHODS FOR ROTA- TION MATRIX LEARNING IN TRAINABLE EMBEDDING INDEXES |
d248476097 | The past few years have witnessed the great success of Diffusion models (DMs) in generating high-fidelity samples in generative modeling tasks. A major limitation of the DM is its notoriously slow sampling procedure which normally requires hundreds to thousands of time discretization steps of the learned diffusion process to reach the desired accuracy. Our goal is to develop a fast sampling method for DMs with fewer steps while retaining high sample quality. To this end, we systematically analyze the sampling procedure in DMs and identify key factors that affect the sample quality, among which the method of discretization is most crucial. By carefully examining the learned diffusion process, we propose Diffusion Exponential Integrator Sampler (DEIS). It is based on the Exponential Integrator designed for discretizing ordinary differential equations (ODEs) and leverages a semilinear structure of the learned diffusion process to reduce the discretization error. The proposed method can be applied to any DMs and can generate highfidelity samples in as few as 10 steps. Moreover, by directly using pre-trained DMs, we achieve state-of-art sampling performance when the number of score function evaluation (NFE) is limited, e.g., 4.17 FID with 10 NFEs, 2.86 FID with only 20 NFEs on CIFAR10. Project page and code: https://qsh-zh.github.io/deis. | FAST SAMPLING OF DIFFUSION MODELS WITH EXPO- NENTIAL INTEGRATOR |
d51747407 | We introduce a new unsupervised representation learning and visualization using deep convolutional networks and self organizing maps called Deep Neural Maps (DNM). DNM jointly learns an embedding of the input data and a mapping from the embedding space to a two-dimensional lattice. We compare visualizations of DNM with those of t-SNE and LLE on the MNIST and COIL-20 data sets. Our experiments show that the DNM can learn efficient representations of the input data, which reflects characteristics of each class. This is shown via backprojecting the neurons of the map on the data space. | Workshop track -ICLR 2018 DEEP NEURAL MAPS |
d1988653 | Stochastic binary hidden units in a multi-layer perceptron (MLP) network give at least three potential benefits when compared to deterministic MLP networks. (1) They allow to learn one-to-many type of mappings.(2) They can be used in structured prediction problems, where modeling the internal structure of the output is important.(3) Stochasticity has been shown to be an excellent regularizer, which makes generalization performance potentially better in general. However, training stochastic networks is considerably more difficult. We study training using M samples of hidden activations per input. We show that the case M = 1 leads to a fundamentally different behavior where the network tries to avoid stochasticity. We propose two new estimators for the training gradient and propose benchmark tests for comparing training algorithms. Our experiments confirm that training stochastic networks is difficult and show that the proposed two estimators perform favorably among all the five known estimators. | Techniques for Learning Binary Stochastic Feedforward Neural Networks |
d235458009 | An important paradigm of natural language processing consists of large-scale pretraining on general domain data and adaptation to particular tasks or domains. As we pre-train larger models, full fine-tuning, which retrains all model parameters, becomes less feasible. Using GPT-3 175B as an example -deploying independent instances of fine-tuned models, each with 175B parameters, is prohibitively expensive. We propose Low-Rank Adaptation, or LoRA, which freezes the pretrained model weights and injects trainable rank decomposition matrices into each layer of the Transformer architecture, greatly reducing the number of trainable parameters for downstream tasks. Compared to GPT-3 175B fine-tuned with Adam, LoRA can reduce the number of trainable parameters by 10,000 times and the GPU memory requirement by 3 times. LoRA performs on-par or better than finetuning in model quality on RoBERTa, DeBERTa, GPT-2, and GPT-3, despite having fewer trainable parameters, a higher training throughput, and, unlike adapters, no additional inference latency. We also provide an empirical investigation into rank-deficiency in language model adaptation, which sheds light on the efficacy of LoRA. We release a package that facilitates the integration of LoRA with PyTorch models and provide our implementations and model checkpoints for RoBERTa, DeBERTa, and GPT-2 at https://github.com/microsoft/LoRA. * Equal contribution. 0 Compared to V1, this draft includes better baselines, experiments on GLUE, and more on adapter latency. 1 While GPT-3 175B achieves non-trivial performance with few-shot learning, fine-tuning boosts its performance significantly as shown in Appendix A. 1 arXiv:2106.09685v2 [cs.CL] 16 Oct 2021 often introduce inference latency (Houlsby et al., 2019; Rebuffi et al., 2017) by extending model depth or reduce the model's usable sequence length (Li & Liang, 2021; Lester et al., 2021; Hambardzumyan et al., 2020; Liu et al., 2021) (Section 3). More importantly, these method often fail to match the fine-tuning baselines, posing a trade-off between efficiency and model quality.We take inspiration from Li et al. (2018a);Aghajanyan et al. (2020)which show that the learned over-parametrized models in fact reside on a low intrinsic dimension. We hypothesize that the change in weights during model adaptation also has a low "intrinsic rank", leading to our proposed Low-Rank Adaptation (LoRA) approach. LoRA allows us to train some dense layers in a neural network indirectly by optimizing rank decomposition matrices of the dense layers' change during adaptation instead, while keeping the pre-trained weights frozen, as shown inFigure 1. Using GPT-3 175B as an example, we show that a very low rank (i.e., r inFigure 1can be one or two) suffices even when the full rank (i.e., d) is as high as 12,288, making LoRA both storage-and compute-efficient.LoRA possesses several key advantages. networks outperform kernel methods? arXiv preprint arXiv: | LORA: LOW-RANK ADAPTATION OF LARGE LAN- GUAGE MODELS |
d211132598 | Federated learning allows edge devices to collaboratively learn a shared model while keeping the training data on device, decoupling the ability to do model training from the need to store the data in the cloud. We propose the Federated matched averaging (FedMA) algorithm designed for federated learning of modern neural network architectures e.g. convolutional neural networks (CNNs) and LSTMs. FedMA constructs the shared global model in a layer-wise manner by matching and averaging hidden elements (i.e. channels for convolution layers; hidden states for LSTM; neurons for fully connected layers) with similar feature extraction signatures. Our experiments indicate that FedMA not only outperforms popular state-of-the-art federated learning algorithms on deep CNN and LSTM architectures trained on real world datasets, but also reduces the overall communication burden. | Published as a conference paper at ICLR 2020 FEDERATED LEARNING WITH MATCHED AVERAGING |
d246867209 | Pruning neural networks at initialization would enable us to find sparse models that retain the accuracy of the original network while consuming fewer computational resources for training and inference. However, current methods are insufficient to enable this optimization and lead to a large degradation in model performance. In this paper, we identify a fundamental limitation in the formulation of current methods, namely that their saliency criteria look at a single step at the start of training without taking into account the trainability of the network. While pruning iteratively and gradually has been shown to improve pruning performance, explicit consideration of the training stage that will immediately follow pruning has so far been absent from the computation of the saliency criterion. To overcome the short-sightedness of existing methods, we propose Prospect Pruning (ProsPr), which uses meta-gradients through the first few steps of optimization to determine which weights to prune. ProsPr combines an estimate of the higherorder effects of pruning on the loss and the optimization trajectory to identify the trainable sub-network. Our method achieves state-of-the-art pruning performance on a variety of vision classification tasks, with less data and in a single shot compared to existing pruning-at-initialization methods. Our code is available online at https://github.com/ | Published as a conference paper at ICLR 2022 PROSPECT PRUNING: FINDING TRAINABLE WEIGHTS AT INITIALIZATION USING META-GRADIENTS |
d21529792 | Reinforcement learning provides a powerful and general framework for decision making and control, but its application in practice is often hindered by the need for extensive feature and reward engineering. Deep reinforcement learning methods can remove the need for explicit engineering of policy or value features, but still require a manually specified reward function. Inverse reinforcement learning holds the promise of automatic reward acquisition, but has proven exceptionally difficult to apply to large, high-dimensional problems with unknown dynamics. In this work, we propose AIRL, a practical and scalable inverse reinforcement learning algorithm based on an adversarial reward learning formulation. We demonstrate that AIRL is able to recover reward functions that are robust to changes in dynamics, enabling us to learn policies even under significant variation in the environment seen during training. Our experiments show that AIRL greatly outperforms prior methods in these transfer settings. | Published as a conference paper at ICLR 2018 LEARNING ROBUST REWARDS WITH ADVERSARIAL INVERSE REINFORCEMENT LEARNING |
d3695872 | This paper proposes a class of well-conditioned neural networks in which a unit amount of change in the inputs causes at most a unit amount of change in the outputs or any of the internal layers. We develop the known methodology of controlling Lipschitz constants to realize its full potential in maximizing robustness, with a new regularization scheme for linear layers, new ways to adapt nonlinearities and a new loss function. With MNIST and CIFAR-10 classifiers, we demonstrate a number of advantages. Without needing any adversarial training, the proposed classifiers exceed the state of the art in robustness against white-box L 2 -bounded adversarial attacks. They generalize better than ordinary networks from noisy data with partially random labels. Their outputs are quantitatively meaningful and indicate levels of confidence and generalization, among other desirable properties. | L 2 -NONEXPANSIVE NEURAL NETWORKS |
d5696027 | Information propagation is a hard task where the goal is to predict users behavior. We introduce an extension of a model which make use of a kernel to modelize diffusion in a latent space. This extension introduce a threhsold to differentiate if users are contaminated or not. | Predict Information Diffusion using a Latent Representation Space Predict Information Diffusion using a Latent Representation Space |
d209531937 | Neural architecture search (NAS) has achieved breakthrough success in a great number of applications in the past few years. It could be time to take a step back and analyze the good and bad aspects in the field of NAS. A variety of algorithms search architectures under different search space. These searched architectures are trained using different setups, e.g., hyper-parameters, data augmentation, regularization. This raises a comparability problem when comparing the performance of various NAS algorithms. NAS-Bench-101 has shown success to alleviate this problem. In this work, we propose an extension to NAS-Bench-101: NAS-Bench-201 with a different search space, results on multiple datasets, and more diagnostic information. NAS-Bench-201 has a fixed search space and provides a unified benchmark for almost any up-to-date NAS algorithms. The design of our search space is inspired from the one used in the most popular cell-based searching algorithms, where a cell is represented as a directed acyclic graph. Each edge here is associated with an operation selected from a predefined operation set. For it to be applicable for all NAS algorithms, the search space defined in NAS-Bench-201 includes all possible architectures generated by 4 nodes and 5 associated operation options, which results in 15,625 neural cell candidates in total. The training log using the same setup and the performance for each architecture candidate are provided for three datasets. This allows researchers to avoid unnecessary repetitive training for selected architecture and focus solely on the search algorithm itself. The training time saved for every architecture also largely improves the efficiency of most NAS algorithms and brings a more computational cost friendly NAS community for a broader range of researchers. We provide additional diagnostic information such as fine-grained loss and accuracy, which can give inspirations to new designs of NAS algorithms. In further support of the proposed NAS-Bench-201, we have analyzed it from many aspects and benchmarked 10 recent NAS algorithms, which verify its applicability.Recently, a variety of NAS algorithms have been increasingly proposed. While these NAS methods are methodically designed and show promising improvements, many setups in their algorithms are * Part of this work was done when Xuanyi was a research intern with Baidu Research. | NAS-BENCH-201: EXTENDING THE SCOPE OF RE- PRODUCIBLE NEURAL ARCHITECTURE SEARCH |
d254221009 | Models trained via empirical risk minimization (ERM) are known to rely on spurious correlations between labels and task-independent input features, resulting in poor generalization to distributional shifts. Group distributionally robust optimization (G-DRO) can alleviate this problem by minimizing the worst-case loss over a set of pre-defined groups over training data. G-DRO successfully improves performance of the worst-group, where the correlation does not hold. However, G-DRO assumes that the spurious correlations and associated worst groups are known in advance, making it challenging to apply it to new tasks with potentially multiple unknown spurious correlations. We propose AGRO-Adversarial Group discovery for Distributionally Robust Optimization-an end-to-end approach that jointly identifies error-prone groups and improves accuracy on them. AGRO equips G-DRO with an adversarial slicing model to find a group assignment for training examples which maximizes worst-case loss over the discovered groups. On the WILDS benchmark, AGRO results in 8% higher model performance on average on known worst-groups, compared to prior group discovery approaches used with G-DRO. AGRO also improves out-of-distribution performance on SST2, QQP, and MS-COCO-datasets where potential spurious correlations are as yet uncharacterized. Human evaluation of ARGO groups shows that they contain well-defined, yet previously unstudied spurious correlations that lead to model errors. | Arxiv Submission AGRO: ADVERSARIAL DISCOVERY OF ERROR-PRONE GROUPS FOR ROBUST OPTIMIZATION |
d232013968 | Autoregressive models are widely used for tasks such as image and audio generation. The sampling process of these models, however, does not allow interruptions and cannot adapt to real-time computational resources. This challenge impedes the deployment of powerful autoregressive models, which involve a slow sampling process that is sequential in nature and typically scales linearly with respect to the data dimension. To address this difficulty, we propose a new family of autoregressive models that enables anytime sampling. Inspired by Principal Component Analysis, we learn a structured representation space where dimensions are ordered based on their importance with respect to reconstruction. Using an autoregressive model in this latent space, we trade off sample quality for computational efficiency by truncating the generation process before decoding into the original data space. Experimentally, we demonstrate in several image and audio generation tasks that sample quality degrades gracefully as we reduce the computational budget for sampling. The approach suffers almost no loss in sample quality (measured by FID) using only 60% to 80% of all latent dimensions for image data. Code is available at https://github.com/Newbeeer/Anytime-Auto-Regressive-Model. | Published as a conference paper at ICLR 2021 ANYTIME SAMPLING FOR AUTOREGRESSIVE MODELS VIA ORDERED AUTOENCODING |
d222398658 | Explainability in AI is crucial for model development, compliance with regulation, and providing operational nuance to predictions.The Shapley framework for explainability attributes a model's predictions to its input features in a mathematically principled and model-agnostic way.However, general implementations of Shapley explainability make an untenable assumption: that the model's features are uncorrelated.In this work, we demonstrate unambiguous drawbacks of this assumption and develop two solutions to Shapley explainability that respect the data manifold.One solution, based on generative modelling, provides flexible access to data imputations; the other directly learns the Shapley value-function, providing performance and stability at the cost of flexibility.While "off-manifold" Shapley values can (i) give rise to incorrect explanations, (ii) hide implicit model dependence on sensitive attributes, and (iii) lead to unintelligible explanations in higher-dimensional data, on-manifold explainability overcomes these problems. | |
d247011732 | Denoising Diffusion Probabilistic Models (DDPMs) can generate high-quality samples such as image and audio samples. However, DDPMs require hundreds to thousands of iterations to produce final samples. Several prior works have successfully accelerated DDPMs through adjusting the variance schedule (e.g., Improved Denoising Diffusion Probabilistic Models) or the denoising equation (e.g., Denoising Diffusion Implicit Models (DDIMs)). However, these acceleration methods cannot maintain the quality of samples and even introduce new noise at a high speedup rate, which limit their practicability. To accelerate the inference process while keeping the sample quality, we provide a fresh perspective that DDPMs should be treated as solving differential equations on manifolds. Under such a perspective, we propose pseudo numerical methods for diffusion models (PNDMs). Specifically, we figure out how to solve differential equations on manifolds and show that DDIMs are simple cases of pseudo numerical methods. We change several classical numerical methods to corresponding pseudo numerical methods and find that the pseudo linear multi-step method is the best in most situations. According to our experiments, by directly using pre-trained models on Cifar10, CelebA and LSUN, PNDMs can generate higher quality synthetic images with only 50 steps compared with 1000-step DDIMs (20x speedup), significantly outperform DDIMs with 250 steps (by around 0.4 in FID) and have good generalization on different variance schedules. | Published as a conference paper at ICLR 2022 PSEUDO NUMERICAL METHODS FOR DIFFUSION MODELS ON MANIFOLDS |
d232404024 | While autoregressive models excel at image compression, their sample quality is often lacking. Although not realistic, generated images often have high likelihood according to the model, resembling the case of adversarial examples. Inspired by a successful adversarial defense method, we incorporate randomized smoothing into autoregressive generative modeling. We first model a smoothed version of the data distribution, and then reverse the smoothing process to recover the original data distribution. This procedure drastically improves the sample quality of existing autoregressive models on several synthetic and real-world image datasets while obtaining competitive likelihoods on synthetic datasets. | Published as a conference paper at ICLR 2021 IMPROVED AUTOREGRESSIVE MODELING WITH DISTRIBUTION SMOOTHING |
d257255456 | Reward design in reinforcement learning (RL) is challenging since specifying human notions of desired behavior may be difficult via reward functions or require many expert demonstrations. Can we instead cheaply design rewards using a natural language interface? This paper explores how to simplify reward design by prompting a large language model (LLM) such as GPT-3 as a proxy reward function, where the user provides a textual prompt containing a few examples (few-shot) or a description (zero-shot) of the desired behavior. Our approach leverages this proxy reward function in an RL framework. Specifically, users specify a prompt once at the beginning of training. During training, the LLM evaluates an RL agent's behavior against the desired behavior described by the prompt and outputs a corresponding reward signal. The RL agent then uses this reward to update its behavior. We evaluate whether our approach can train agents aligned with user objectives in the Ultimatum Game, matrix games, and the DEALORNODEAL negotiation task. In all three tasks, we show that RL agents trained with our framework are well-aligned with the user's objectives and outperform RL agents trained with reward functions learned via supervised learning. Code and prompts can be found here. | Published as a conference paper at ICLR 2023 REWARD DESIGN WITH LANGUAGE MODELS |
d68071305 | We propose Generative Predecessor Models for Imitation Learning (GPRIL), a novel imitation learning algorithm that matches the state-action distribution to the distribution observed in expert demonstrations, using generative models to reason probabilistically about alternative histories of demonstrated states. We show that this approach allows an agent to learn robust policies using only a small number of expert demonstrations and self-supervised interactions with the environment. We derive this approach from first principles and compare it empirically to a stateof-the-art imitation learning method, showing that it outperforms or matches its performance on two simulated robot manipulation tasks and demonstrate significantly higher sample efficiency by applying the algorithm on a real robot. Recent advances in generative modeling, such as Goodfellow et al. (2014); Kingma and Welling (2013); Van Den Oord et al. (2016b;a); Dinh et al. (2016), have shown great promise at modeling complex distributions and can be used to reason probabilistically about such state-action pairs. * This work was carried out at DeepMind. | GENERATIVE PREDECESSOR MODELS FOR SAMPLE- EFFICIENT IMITATION LEARNING |
d231847288 | Despite definite success in deep reinforcement learning problems, actor-critic algorithms are still confronted with sample inefficiency in complex environments, particularly in tasks where efficient exploration is a bottleneck. These methods consider a policy (the actor) and a value function (the critic) whose respective losses are built using different motivations and approaches. This paper introduces a third protagonist: the adversary. While the adversary mimics the actor by minimizing the KL-divergence between their respective action distributions, the actor, in addition to learning to solve the task, tries to differentiate itself from the adversary predictions. This novel objective stimulates the actor to follow strategies that could not have been correctly predicted from previous trajectories, making its behavior innovative in tasks where the reward is extremely rare. Our experimental analysis shows that the resulting Adversarially Guided Actor-Critic (AGAC) algorithm leads to more exhaustive exploration. Notably, AGAC outperforms current state-of-the-art methods on a set of various hard-exploration and procedurally-generated tasks. * Equal contribution.Published as a conference paper at ICLR 2021 This paper analyses and explores how AGAC explicitly drives diversity in the behaviors of the agent while remaining reward-focused, and to which extent this approach allows to adapt to the evolving state space of procedurally-generated environments where the map is constructed differently with each new episode. Moreover, because stability is a legitimate concern since specific instances of adversarial networks were shown to be prone to hyperparameter sensitivity issues(Arjovsky & Bottou, 2017), we also examine this aspect in our experiments.The contributions of this work are as follow: (i) we propose a novel actor-critic formulation inspired from adversarial learning (AGAC), (ii) we analyse empirically AGAC on key reinforcement learning aspects such as diversity, exploration and stability, (iii) we demonstrate significant gains in performance on several sparse-reward hard-exploration tasks including procedurally-generated tasks. | Published as a conference paper at ICLR 2021 ADVERSARIALLY GUIDED ACTOR-CRITIC |
d54444711 | We view molecular optimization as a graph-to-graph translation problem. The goal is to learn to map from one molecular graph to another with better properties based on an available corpus of paired molecules. Since molecules can be optimized in different ways, there are multiple viable translations for each input graph. A key challenge is therefore to model diverse translation outputs. Our primary contributions include a junction tree encoder-decoder for learning diverse graph translations along with a novel adversarial training method for aligning distributions of molecules. Diverse output distributions in our model are explicitly realized by low-dimensional latent vectors that modulate the translation process. We evaluate our model on multiple molecular optimization tasks and show that our model outperforms previous state-of-the-art baselines by a significant margin. | LEARNING MULTIMODAL GRAPH-TO-GRAPH TRANS- LATION FOR MOLECULAR OPTIMIZATION |
d254246465 | Empirical studies suggest that machine learning models trained with empirical risk minimization (ERM) often rely on attributes that may be spuriously correlated with the class labels. Such models typically lead to poor performance during inference for data lacking such correlations. In this work, we explicitly consider a situation where potential spurious correlations are present in the majority of training data. In contrast with existing approaches, which use the ERM model outputs to detect the samples without spurious correlations and either heuristically upweight or upsample those samples, we propose the logit correction (LC) loss, a simple yet effective improvement on the softmax cross-entropy loss, to correct the sample logit. We demonstrate that minimizing the LC loss is equivalent to maximizing the group-balanced accuracy, so the proposed LC could mitigate the negative impacts of spurious correlations. Our extensive experimental results further reveal that the proposed LC loss outperforms state-of-the-art solutions on multiple popular benchmarks by a large margin, an average 5.5% absolute improvement, without access to spurious attribute labels. LC is also competitive with oracle methods that make use of the attribute labels. † * Work primarily done during an internship at Amazon. † Code is available at https | Published as a conference paper at ICLR 2023 AVOIDING SPURIOUS CORRELATIONS VIA LOGIT COR- RECTION |
d231648272 | Exploration under sparse reward is a long-standing challenge of model-free reinforcement learning. The state-of-the-art methods address this challenge by introducing intrinsic rewards to encourage exploration in novel states or uncertain environment dynamics. Unfortunately, methods based on intrinsic rewards often fall short in procedurally-generated environments, where a different environment is generated in each episode so that the agent is not likely to visit the same state more than once. Motivated by how humans distinguish good exploration behaviors by looking into the entire episode, we introduce RAPID, a simple yet effective episode-level exploration method for procedurally-generated environments. RAPID regards each episode as a whole and gives an episodic exploration score from both per-episode and long-term views. Those highly scored episodes are treated as good exploration behaviors and are stored in a small ranking buffer. The agent then imitates the episodes in the buffer to reproduce the past good exploration behaviors. We demonstrate our method on several procedurally-generated MiniGrid environments, a first-person-view 3D Maze navigation task from MiniWorld, and several sparse MuJoCo tasks. The results show that RAPID significantly outperforms the state-of-the-art intrinsic reward strategies in terms of sample efficiency and final performance. The code is available at | Published as a conference paper at ICLR 2021 RANK THE EPISODES: A SIMPLE APPROACH FOR EXPLORATION IN PROCEDURALLY-GENERATED ENVIRONMENTS |
d252668844 | Neural networks embed the geometric structure of a data manifold lying in a high-dimensional space into latent representations. Ideally, the distribution of the data points in the latent space should depend only on the task, the data, the loss, and other architecture-specific constraints. However, factors such as the random weights initialization, training hyperparameters, or other sources of randomness in the training phase may induce incoherent latent spaces that hinder any form of reuse. Nevertheless, we empirically observe that, under the same data and modeling choices, the angles between the encodings within distinct latent spaces do not change. In this work, we propose the latent similarity between each sample and a fixed set of anchors as an alternative data representation, demonstrating that it can enforce the desired invariances without any additional training. We show how neural architectures can leverage these relative representations to guarantee, in practice, invariance to latent isometries and rescalings, effectively enabling latent space communication: from zero-shot model stitching to latent space comparison between diverse settings. We extensively validate the generalization capability of our approach on different datasets, spanning various modalities ( | Published as a conference paper at ICLR 2023 RELATIVE REPRESENTATIONS ENABLE ZERO-SHOT LATENT SPACE COMMUNICATION |
d212633559 | Neural Ordinary Differential Equations (NODEs) have proven to be a powerful modeling tool for approximating (interpolation) and forecasting (extrapolation) irregularly sampled time series data. However, their performance degrades substantially when applied to real-world data, especially long-term data with complex behaviors (e.g., long-term trend across years, mid-term seasonality across months, and short-term local variation across days). To address the modeling of such complex data with different behaviors at different frequencies (time spans), we propose a novel progressive learning paradigm of NODEs for long-term time series forecasting. Specifically, following the principle of curriculum learning, we gradually increase the complexity of data and network capacity as training progresses. Our experiments with both synthetic data and real traffic data (PeMS Bay Area traffic data) show that our training methodology consistently improves the performance of vanilla NODEs by over 64%. | PROGRESSIVE GROWING OF NEURAL ODES |
d3347806 | We propose a neural language model capable of unsupervised syntactic structure induction. The model leverages the structure information to form better semantic representations and better language modeling. Standard recurrent neural networks are limited by their structure and fail to efficiently use syntactic information. On the other hand, tree-structured recursive networks usually require additional structural supervision at the cost of human expert annotation. In this paper, We propose a novel neural language model, called the Parsing-Reading-Predict Networks (PRPN), that can simultaneously induce the syntactic structure from unannotated sentences and leverage the inferred structure to learn a better language model. In our model, the gradient can be directly back-propagated from the language model loss into the neural parsing network. Experiments show that the proposed model can discover the underlying syntactic structure and achieve state-of-the-art performance on word/character-level language model tasks. | Under review as a conference paper NEURAL LANGUAGE MODELING BY JOINTLY LEARNING SYNTAX AND LEXICON |
d247793212 | Existing continual learning methods use Batch Normalization (BN) to facilitate training and improve generalization across tasks. However, the non-i.i.d and nonstationary nature of continual learning data, especially in the online setting, amplify the discrepancy between training and testing in BN and hinder the performance of older tasks. In this work, we study the cross-task normalization effect of BN in online continual learning where BN normalizes the testing data using moments biased towards the current task, resulting in higher catastrophic forgetting. This limitation motivates us to propose a simple yet effective method that we call Continual Normalization (CN) to facilitate training similar to BN while mitigating its negative effect. Extensive experiments on different continual learning algorithms and online scenarios show that CN is a direct replacement for BN and can provide substantial performance improvements. Our implementation is available at https://github.com/phquang/Continual-Normalization. | Published as a conference paper at ICLR 2022 CONTINUAL NORMALIZATION: RETHINKING BATCH NORMALIZATION FOR ONLINE CONTINUAL LEARNING |
d257496586 | Due to its geometric properties, hyperbolic space can support high-fidelity embeddings of tree-and graph-structured data, upon which various hyperbolic networks have been developed. Existing hyperbolic networks encode geometric priors not only for the input, but also at every layer of the network. This approach involves repeatedly mapping to and from hyperbolic space, which makes these networks complicated to implement, computationally expensive to scale, and numerically unstable to train. In this paper, we propose a simpler approach: learn a hyperbolic embedding of the input, then map once from it to Euclidean space using a mapping that encodes geometric priors by respecting the isometries of hyperbolic space, and finish with a standard Euclidean network. The key insight is to use a random feature mapping via the eigenfunctions of the Laplace operator, which we show can approximate any isometry-invariant kernel on hyperbolic space. Our method can be used together with any graph neural networks: using even a linear graph model yields significant improvements in both efficiency and performance over other hyperbolic baselines in both transductive and inductive tasks. Hjelm. Deep graph infomax. ICLR (Poster), 2(3):4, 2019.Ming-Xi Wang. Laplacian eigenspaces, horocycles and neuron models on hyperbolic spaces. 2020. | Published as a conference paper at ICLR 2023 RANDOM LAPLACIAN FEATURES FOR LEARNING WITH HYPERBOLIC SPACE |
d3566136 | Common-sense physical reasoning is an essential ingredient for any intelligent agent operating in the real-world. For example, it can be used to simulate the environment, or to infer the state of parts of the world that are currently unobserved. In order to match real-world conditions this causal knowledge must be learned without access to supervised data. To address this problem we present a novel method that learns to discover objects and model their physical interactions from raw visual images in a purely unsupervised fashion. It incorporates prior knowledge about the compositional nature of human perception to factor interactions between object-pairs and learn efficiently. On videos of bouncing balls we show the superior modelling capabilities of our method compared to other unsupervised neural approaches that do not incorporate such prior knowledge. We demonstrate its ability to handle occlusion and show that it can extrapolate learned knowledge to scenes with different numbers of objects. * Work performed while at IDSIA. | Published as a conference paper at ICLR 2018 RELATIONAL NEURAL EXPECTATION MAXIMIZATION: UNSUPERVISED DISCOVERY OF OBJECTS AND THEIR INTERACTIONS |
d14124313 | In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve the stateof-the-art results. Importantly, we have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.1 | Very Deep Convolutional Networks for Large-Scale Image Recognition |
d8328889 | We present an efficient document representation learning framework, Document Vector through Corruption (Doc2VecC). Doc2VecC represents each document as a simple average of word embeddings. It ensures a representation generated as such captures the semantic meanings of the document during learning. A corruption model is included, which introduces a data-dependent regularization that favors informative or rare words while forcing the embeddings of common and non-discriminative ones to be close to zero. Doc2VecC produces significantly better word embeddings than Word2Vec. We compare Doc2VecC with several state-of-the-art document representation learning algorithms. The simple model architecture introduced by Doc2VecC matches or out-performs the state-of-the-art in generating high-quality document representations for sentiment analysis, document classification as well as semantic relatedness tasks. The simplicity of the model enables training on billions of words per hour on a single machine. At the same time, the model is very efficient in generating representations of unseen documents at test time. | Published as a conference paper at ICLR 2017 EFFICIENT VECTOR REPRESENTATION FOR DOCU- MENTS THROUGH CORRUPTION |
d202677369 | We propose a black-box algorithm called Adversarial Variational Inference and Learning (AdVIL) to perform inference and learning in a general Markov random field (MRF). AdVIL employs two variational distributions to approximately infer the latent variables and estimate the partition function of an MRF, respectively. The two variational distributions provide an estimate of the negative log-likelihood of the MRF as a minimax optimization problem, which is solved by stochastic gradient descent. AdVIL is proven convergent under certain conditions. On one hand, compared to the contrastive divergence, AdVIL requires minimal assumptions about the model structure and can deal with a broader family of MRFs. On the other hand, compared to existing black-box methods, AdVIL provides a tighter estimate of the log partition function and achieves much better empirical results. * | Published as a conference paper at ICLR 2020 TO RELIEVE YOUR HEADACHE OF TRAINING AN MRF, TAKE ADVIL |
d251135247 | Deep neural networks (DNNs) often have to be compressed, via pruning and/or quantization, before they can be deployed in practical settings. In this work we propose a new compression-aware minimizer dubbed CrAM that modifies the optimization step in a principled way, in order to produce models whose local loss behavior is stable under compression operations such as pruning. Thus, dense models trained via CrAM should be compressible post-training, in a single step, without significant accuracy loss. Experimental results on standard benchmarks, such as residual networks for ImageNet classification and BERT models for language modelling, show that CrAM produces dense models that can be more accurate than the standard SGD/Adam-based baselines, but which are stable under weight pruning: specifically, we can prune models in one-shot to 70-80% sparsity with almost no accuracy loss, and to 90% with reasonable (∼ 1%) accuracy loss, which is competitive with gradual compression methods. Additionally, CrAM can produce sparse models which perform well for transfer learning, and it also works for semi-structured 2:4 pruning patterns supported by GPU hardware. The code for reproducing the results is available at: https://github.com/IST-DASLab/CrAM.Published as a conference paper at ICLR 2023 provide a well-performing default value. We detail the CrAM algorithm and provide a theoretical motivation leveraging fundamental results in robust optimization(Danskin, 2012)in Section 3.To complement our algorithmic contribution, we perform an extensive experimental analysis of CrAM. We mainly focus on compression via weight pruning, but we also show that CrAM is compatible with weight quantization. Generally, CrAM models trained on large-scale image classification or language modelling tasks can improve over the dense baseline performance, while being very robust to one-shot pruning, at different sparsity levels. For image classification, CrAM can train a highly-accurate dense ResNet50 model on ImageNet, that can be pruned in one-shot to 80% and 90% sparsity, and is competitive in terms of accuracy relative to state-of-the-art gradual pruning methods, following an inexpensive Batch Normalization re-tuning step on a small calibration set.Moreover, we show that full CrAM training is not necessary for good performance: specifically, a short CrAM finetuning period is sufficient to substantially improve one-shot pruning accuracy. For instance, using CrAM to transfer the standard BERT-base model (Devlin et al., 2019) on SQuADv1.1 question-answering (Rajpurkar et al., 2016), we obtain models that are both more accurate and more compressible than with optimizers such as Adam (Kingma & Ba, 2015) or SAM (Foret et al., 2021). In addition, a short (≤ 2 epochs) finetuning of the sparse model can provide substantial additional improvements: the 80%-sparse CrAM finetuned model reaches higher accuracy than the highlycompetitive gradual pruning methods PLATON (Zhang et al., 2022) and Movement Pruning(Sanh et al., 2020), at a fraction of the training budget.CrAM lends itself to several extensions: it can be used with different layer-wise sparsity distributions, semi-structured N:M sparsity patterns, and one-shot pruning techniques. Sparse CrAM models can be successfully used for sparse transfer learning, where they can perform well on a wide range of "downstream" target tasks, even when compared to pruning methods that train a separate model for each sparsity level. We also provide evidence that the CrAM update can produce models that are robust to quantization. Similar to SAM (Foret et al., 2021), one limitation is the added computational cost, as CrAM requires an additional backwards pass for the model perturbation. This can be addressed by only performing limited finetuning via CrAM instead of full retraining, or by only performing a regular optimization step for a fraction of the time, both of which we show to have a limited impact on accuracy. Moreover, our approach is also compatible with efficient SAM-type updates(Liu et al., 2020;Du et al., 2022a). We also provide a well-performing variant of CrAM that uses sparse gradients, which could be leveraged by frameworks with support for sparse back-propagation (Nikdan et al., 2023). | Published as a conference paper at ICLR 2023 CRAM: A COMPRESSION-AWARE MINIMIZER |
d222341795 | Machine learning models have traditionally been developed under the assumption that the training and test distributions match exactly. However, recent success in few-shot learning and related problems are encouraging signs that these models can be adapted to more realistic settings where train and test distributions differ. Unfortunately, there is severely limited theoretical support for these algorithms and little is known about the difficulty of these problems. In this work, we provide novel information-theoretic lower-bounds on minimax rates of convergence for algorithms that are trained on data from multiple sources and tested on novel data. Our bounds depend intuitively on the information shared between sources of data, and characterize the difficulty of learning in this setting for arbitrary algorithms. We demonstrate these bounds on a hierarchical Bayesian model of meta-learning, computing both upper and lower bounds on parameter estimation via maximum-a-posteriori inference.1 arXiv:2010.07140v1 [stat.ML] 14 Oct 20201. Note that this definition encompasses few-shot learning. 1≤j≤J ρ(θ W (Z), θ(P j )) Using Markov's inequality, and then following the proof of Theorem 4, we have,Proof This result follows as an application of the data processing inequality. Notice that π M +1 → π 1:M → W forms a Markov chain. Thus,by the data processing inequality. We can compute I(π M +1 ; π 1:M ) in closed form:The proof is completed by plugging in the i.i.d local packing bound alongside the above.B.3. Local packing resultsLemma 11 (Meta-learning local packing) Consider the same setting as in Theorem 1, thenProof There are (J − 1)!/(J − M − 1)! orderings on the first M indices, given the (M + 1) th . We introduce the following notation,As in previous proofs, we notice that we can write,First, note that we can upper bound I(π M +1 ; W ) ≤ nI(π M +1 ; w), where w denotes a single row in W . Further, | Theoretical bounds on estimation error for meta-learning Mengye Ren African Master for Mathematical Sciences; Vector Institute |
d6230637 | In this paper we present a method for learning a discriminative classifier from unlabeled or partially labeled data. Our approach is based on an objective function that trades-off mutual information between observed examples and their predicted categorical class distribution, against robustness of the classifier to an adversarial generative model. The resulting algorithm can either be interpreted as a natural generalization of the generative adversarial networks (GAN) framework or as an extension of the regularized information maximization (RIM) framework to robust classification against an optimal adversary. We empirically evaluate our method -which we dub categorical generative adversarial networks (or CatGAN) -on synthetic data as well as on challenging image classification tasks, demonstrating the robustness of the learned classifiers. We further qualitatively assess the fidelity of samples generated by the adversarial generator that is learned alongside the discriminative classifier, and identify links between the CatGAN objective and discriminative clustering algorithms (such as RIM). | Published as a conference paper at ICLR 2016 UNSUPERVISED AND SEMI-SUPERVISED LEARNING WITH CATEGORICAL GENERATIVE ADVERSARIAL NETWORKS |
d246904340 | Transformer has shown great successes in natural language processing, computer vision, and audio processing. As one of its core components, the softmax attention helps to capture long-range dependencies yet prohibits its scale-up due to the quadratic space and time complexity to the sequence length. Kernel methods are often adopted to reduce the complexity by approximating the softmax operator. Nevertheless, due to the approximation errors, their performances vary in different tasks/corpus and suffer crucial performance drops when compared with the vanilla softmax attention. In this paper, we propose a linear transformer called COSFORMER that can achieve comparable or better accuracy to the vanilla transformer in both casual and cross attentions. COSFORMER is based on two key properties of softmax attention: i). non-negativeness of the attention matrix; ii). a non-linear re-weighting scheme that can concentrate the distribution of the attention matrix. As its linear substitute, COSFORMER fulfills these properties with a linear operator and a cosine-based distance re-weighting mechanism. Extensive experiments on language modeling and text understanding tasks demonstrate the effectiveness of our method. We further examine our method on long sequences and achieve state-of-the-art performance on the Long-Range Arena benchmark. The source code is available at COSFORMER . * Indicates the corresponding author. † Indicates equal contribution.Published as a conference paper at ICLR 2022 recurrent(Hochreiter & Schmidhuber, 1997)and convolutional architectures(He et al., 2016), transformer-based architectures are generally more scalable to data volumes (Brown et al., 2020) and stronger in capturing global information with less inductive bias, thus excelling on many tasks. | Published as a conference paper at ICLR 2022 COSFORMER : RETHINKING SOFTMAX IN ATTENTION |
d257496457 | In this paper, we study a novel inference paradigm, termed as schema inference, that learns to deductively infer the explainable predictions by rebuilding the prior deep neural network (DNN) forwarding scheme, guided by the prevalent philosophical cognitive concept of schema. We strive to reformulate the conventional model inference pipeline into a graph matching policy that associates the extracted visual concepts of an image with the pre-computed scene impression, by analogy with human reasoning mechanism via impression matching. To this end, we devise an elaborated architecture, termed as SchemaNet, as a dedicated instantiation of the proposed schema inference concept, that models both the visual semantics of input instances and the learned abstract imaginations of target categories as topological relational graphs. Meanwhile, to capture and leverage the compositional contributions of visual semantics in a global view, we also introduce a universal Feat2Graph scheme in SchemaNet to establish the relational graphs that contain abundant interaction information. Both the theoretical analysis and the experimental results on several benchmarks demonstrate that the proposed schema inference achieves encouraging performance and meanwhile yields a clear picture of the deductive process leading to the predictions. Our code is available at https://github.com. Model doctor: A simple gradient aggregation strategy for diagnosing and treating cnn classifiers. In | Published as a conference paper at ICLR 2023 SCHEMA INFERENCE FOR INTERPRETABLE IMAGE CLASSIFICATION |
d247291992 | Despite the tremendous empirical success of deep learning models to solve various learning tasks, our theoretical understanding of their generalization ability is very limited. Classical generalization bounds based on tools such as the VC dimension or Rademacher complexity, are so far unsuitable for deep models and it is doubtful that these techniques can yield tight bounds even in the most idealistic settings(Nagarajan & Kolter, 2019). In this work, we instead revisit the concept of leave-one-out (LOO) error to measure the generalization ability of deep models in the so-called kernel regime. While popular in statistics, the LOO error has been largely overlooked in the context of deep learning. By building upon the recently established connection between neural networks and kernel learning, we leverage the closed-form expression for the leave-one-out error, giving us access to an efficient proxy for the test error. We show both theoretically and empirically that the leave-one-out error is capable of capturing various phenomena in generalization theory, such as double descent, random labels or transfer learning. Our work therefore demonstrates that the leave-one-out error provides a tractable way to estimate the generalization ability of deep neural networks in the kernel regime, opening the door to potential, new research directions in the field of generalization. | Published as a conference paper at ICLR 2022 GENERALIZATION THROUGH THE LENS OF LEAVE- ONE-OUT ERROR |
d15323440 | The Weyl transform is introduced as a powerful framework for representing measurement data. Transform coefficients are connected to the Walsh-Hadamard transform of multiscale autocorrelations, and different forms of dyadic periodicity in a signal are shown to appear as different features in its Weyl coefficients. A large group of multiscale transformations is shown to support very fast pooling since the Weyl coefficients are unique up to permutation and phase changes when the original signal is transformed by any element of this group. The effectiveness of the Weyl transform is demonstrated through the example of textured image classification. * Q. Qiu and A. Thompson contributed equally to this work. | REPRESENTATION USING THE WEYL TRANSFORM |
d251402961 | We present Bit Diffusion: a simple and generic approach for generating discrete data with continuous state and continuous time diffusion models. The main idea behind our approach is to first represent the discrete data as binary bits, and then train a continuous diffusion model to model these bits as real numbers which we call analog bits. To generate samples, the model first generates the analog bits, which are then thresholded to obtain the bits that represent the discrete variables. We further propose two simple techniques, namely Self-Conditioning and Asymmetric Time Intervals, which lead to a significant improvement in sample quality. Despite its simplicity, the proposed approach can achieve strong performance in both discrete image generation and image captioning tasks. For discrete/categorical image generation, we significantly improve previous state-of-the-art on both CIFAR-10 (which has 3K discrete 8-bit tokens) and IMAGENET 64×64 (which has 12K discrete 8-bit tokens), outperforming the best autoregressive model in both sample quality (measured by FID) and efficiency. For image captioning on MS-COCO dataset, our approach achieves competitive results compared to autoregressive models. | Published as a conference paper at ICLR 2023 ANALOG BITS: GENERATING DISCRETE DATA USING DIFFUSION MODELS WITH SELF-CONDITIONING |
d4302773 | The Madry Lab recently hosted a competition designed to test the robustness of their adversarially trained MNIST model. Attacks were constrained to perturb each pixel of the input image by a scaled maximal L ∞ distortion = 0.3. This decision discourages the use of attacks which are not optimized on the L ∞ distortion metric. Our experimental results demonstrate that by relaxing the L ∞ constraint of the competition, the elastic-net attack to deep neural networks (EAD) can generate transferable adversarial examples which, despite their high average L ∞ distortion, have minimal visual distortion. These results call into question the use of L ∞ as a sole measure for visual distortion, and further demonstrate the power of EAD at generating robust adversarial examples. | Workshop track -ICLR 2018 ATTACKING THE MADRY DEFENSE MODEL WITH L 1 -BASED ADVERSARIAL EXAMPLES |
d247451267 | Disentangled representation learning is one of the major goals of deep learning, and is a key step for achieving explainable and generalizable models. A well-defined theoretical guarantee still lacks for the VAE-based unsupervised methods, which are a set of popular methods to achieve unsupervised disentanglement. The Group Theory based definition of representation disentanglement mathematically connects the data transformations to the representations using the formalism of group. In this paper, built on the group-based definition and inspired by the n-th dihedral group, we first propose a theoretical framework towards achieving unsupervised representation disentanglement. We then propose a model, based on existing VAEbased methods, to tackle the unsupervised learning problem of the framework. In the theoretical framework, we prove three sufficient conditions on model, group structure, and data respectively in an effort to achieve, in an unsupervised way, disentangled representation per group-based definition. With the first two of the conditions satisfied and a necessary condition derived for the third one, we offer additional constraints, from the perspective of the group-based definition, for the existing VAE-based models. Experimentally, we train 1800 models covering the most prominent VAE-based methods on five datasets to verify the effectiveness of our theoretical framework. Compared to the original VAE-based methods, these Groupified VAEs consistently achieve better mean performance with smaller variances. | TOWARDS BUILDING A GROUP-BASED UNSUPER- VISED REPRESENTATION DISENTANGLEMENT FRAME- WORK |
d249240147 | Byzantine-robustness has been gaining a lot of attention due to the growth of the interest in collaborative and federated learning. However, many fruitful directions, such as the usage of variance reduction for achieving robustness and communication compression for reducing communication costs, remain weakly explored in the field. This work addresses this gap and proposes Byz-VR-MARINA-a new Byzantine-tolerant method with variance reduction and compression. A key message of our paper is that variance reduction is key to fighting Byzantine workers more effectively. At the same time, communication compression is a bonus that makes the process more communication efficient. We derive theoretical convergence guarantees for Byz-VR-MARINA outperforming previous state-of-the-art for general non-convex and Polyak-Łojasiewicz loss functions. Unlike the concurrent Byzantine-robust methods with variance reduction and/or compression, our complexity results are tight and do not rely on restrictive assumptions such as boundedness of the gradients or limited compression. Moreover, we provide the first analysis of a Byzantine-tolerant method supporting non-uniform sampling of stochastic gradients. Numerical experiments corroborate our theoretical findings. : A novel method for machine learning problems using stochastic recursive | VARIANCE REDUCTION IS AN ANTIDOTE TO BYZANTINE WORKERS: BETTER RATES, WEAKER ASSUMPTIONS AND COMMUNICATION COMPRESSION AS A CHERRY ON THE TOP |
d4410570 | Recurrent neural networks (RNNs) form an important class of architectures among neural networks useful for language modeling and sequential prediction. However, optimizing RNNs is known to be harder compared to feed-forward neural networks. A number of techniques have been proposed in literature to address this problem. In this paper we propose a simple technique called fraternal dropout that takes advantage of dropout to achieve this goal. Specifically, we propose to train two identical copies of an RNN (that share parameters) with different dropout masks while minimizing the difference between their (pre-softmax) predictions. In this way our regularization encourages the representations of RNNs to be invariant to dropout mask, thus being robust. We show that our regularization term is upper bounded by the expectation-linear dropout objective which has been shown to address the gap due to the difference between the train and inference phases of dropout. We evaluate our model and achieve state-of-the-art results in sequence modeling tasks on two benchmark datasets -Penn Treebank and Wikitext-2. We also show that our approach leads to performance improvement by a significant margin in image captioning (Microsoft COCO) and semi-supervised (CIFAR-10) tasks.Published as a conference paper at ICLR 2018Merity et al. (2017a)andMerity et al. (2017b)on the other hand show that activity regularization (AR) and temporal activation regularization (TAR) 1 are also effective methods for regularizing LSTMs. Another more recent way of regularizing RNNs, that is similar in spirit to the approach we take, involves minimizing the difference between the hidden states of the original and the auxiliary networkSerdyuk et al. (2017). | Published as a conference paper at ICLR 2018 FRATERNAL DROPOUT |
d11222874 | Hypothesis testing is an important cognitive process that supports human reasoning. In this paper, we introduce a new computational hypothesis testing framework that is based on memory augmented neural networks. Our approach involves a hypothesis testing loop that reconsiders and progressively refines a previously formed hypothesis in order to generate new hypotheses to test. We applied the proposed approach to language comprehension task by using Neural Semantic Encoders (NSE). Our NSE models achieved the state-of-the-art results showing an absolute improvement of 1.2% to 2.6% accuracy over previous results obtained by single and ensemble systems on standard machine comprehension benchmarks such as the Children's Book Test (CBT) and Who-Did-What (WDW) news article datasets. | Published as a conference paper at ICLR 2017 REASONING WITH MEMORY AUGMENTED NEURAL NETWORKS FOR LANGUAGE COMPREHENSION |
d198895601 | We propose a novel method for unsupervised image-to-image translation, which incorporates a new attention module and a new learnable normalization function in an end-to-end manner. The attention module guides our model to focus on more important regions distinguishing between source and target domains based on the attention map obtained by the auxiliary classifier. Unlike previous attention-based method which cannot handle the geometric changes between domains, our model can translate both images requiring holistic changes and images requiring large shape changes. Moreover, our new AdaLIN (Adaptive Layer-Instance Normalization) function helps our attention-guided model to flexibly control the amount of change in shape and texture by learned parameters depending on datasets. Experimental results show the superiority of the proposed method compared to the existing state-of-the-art models with a fixed network architecture and hyper-parameters. Our code and datasets are available at https | U-GAT-IT: UNSUPERVISED GENERATIVE ATTEN- TIONAL NETWORKS WITH ADAPTIVE LAYER- INSTANCE NORMALIZATION FOR IMAGE-TO-IMAGE TRANSLATION |
d7179166 | Artificial neural networks typically have a fixed, non-linear activation function at each neuron. We have designed a novel form of piecewise linear activation function that is learned independently for each neuron using gradient descent. With this adaptive activation function, we are able to improve upon deep neural network architectures composed of static rectified linear units, achieving state-of-theart performance on CIFAR-10 (7.51%), CIFAR-100 (30.83%), and a benchmark from high-energy physics involving Higgs boson decay modes. | LEARNING ACTIVATION FUNCTIONS TO IMPROVE DEEP NEURAL NETWORKS |
d250089350 | Reward-free reinforcement learning (RF-RL), a recently introduced RL paradigm, relies on random action-taking to explore the unknown environment without any reward feedback information. While the primary goal of the exploration phase in RF-RL is to reduce the uncertainty in the estimated model with minimum number of trajectories, in practice, the agent often needs to abide by certain safety constraint at the same time. It remains unclear how such safe exploration requirement would affect the corresponding sample complexity in order to achieve the desired optimality of the obtained policy in planning. In this work, we make a first attempt to answer this question. In particular, we consider the scenario where a safe baseline policy is known beforehand, and propose a unified Safe reWard-frEe ExploraTion (SWEET) framework. We then particularize the SWEET framework to the tabular and the low-rank MDP settings, and develop algorithms coined Tabular-SWEET and Low-rank-SWEET, respectively. Both algorithms leverage the concavity and continuity of the newly introduced truncated value functions, and are guaranteed to achieve zero constraint violation during exploration with high probability. Furthermore, both algorithms can provably find a near-optimal policy subject to any constraint in the planning phase. Remarkably, the sample complexities under both algorithms match or even outperform the state of the art in their constraint-free counterparts up to some constant factors, proving that safety constraint hardly increases the sample complexity for RF-RL. | SAFE EXPLORATION INCURS NEARLY NO ADDI- TIONAL SAMPLE COMPLEXITY FOR REWARD-FREE RL |
d248405706 | Cross-entropy loss and focal loss are the most common choices when training deep neural networks for classification problems. Generally speaking, however, a good loss function can take on much more flexible forms, and should be tailored for different tasks and datasets. Motivated by how functions can be approximated via Taylor expansion, we propose a simple framework, named PolyLoss, to view and design loss functions as a linear combination of polynomial functions. Our PolyLoss allows the importance of different polynomial bases to be easily adjusted depending on the targeting tasks and datasets, while naturally subsuming the aforementioned cross-entropy loss and focal loss as special cases. Extensive experimental results show that the optimal choice within the PolyLoss is indeed dependent on the task and dataset. Simply by introducing one extra hyperparameter and adding one line of code, our Poly-1 formulation outperforms the crossentropy loss and focal loss on 2D image classification, instance segmentation, object detection, and 3D object detection tasks, sometimes by a large margin. Task ImageNet classification COCO det. and seg. Waymo Open Dataset 3D detection Default loss Cross-entropy Cross-entropy Focal loss arXiv:2204.12511v2 [cs.CV] 10 May 2022Published as a conference paper at ICLR 2022Our study shows that, in order to achieve better results, it is necessary to adjust polynomial coefficients α j for different tasks and datasets. Since it is impossible to adjust an infinite number of α j , we explore various strategies with a small degree of freedom. Perhaps surprisingly, we observe that simply adjusting the single polynomial coefficient for the leading polynomial, which we denote L Poly-1 , is sufficient to achieve significant improvements over the commonly used cross-entropy loss and focal loss. Overall, our contribution can be summarized as: | POLYLOSS: A POLYNOMIAL EXPANSION PERSPEC- TIVE OF CLASSIFICATION LOSS FUNCTIONS |
d237420771 | A Learning hierarchical structures in sequential data-from simple algorithmic patterns to natural language-in a reliable, generalizable way remains a challenging problem for neural language models. Past work has shown that recurrent neural networks (RNNs) struggle to generalize on held-out algorithmic or syntactic patterns without supervision or some inductive bias. To remedy this, many papers have explored augmenting RNNs with various differentiable stacks, by analogy with finite automata and pushdown automata (PDAs). In this paper, we improve the performance of our recently proposed Nondeterministic Stack RNN (NS-RNN), which uses a differentiable data structure that simulates a nondeterministic PDA, with two important changes. First, the model now assigns unnormalized positive weights instead of probabilities to stack actions, and we provide an analysis of why this improves training. Second, the model can directly observe the state of the underlying PDA. Our model achieves lower cross-entropy than all previous stack RNNs on five context-free language modeling tasks (within 0.05 nats of the information-theoretic lower bound), including a task on which the NS-RNN previously failed to outperform a deterministic stack RNN baseline. Finally, we propose a restricted version of the NS-RNN that incrementally processes infinitely long sequences, and we present language modeling results on the Penn Treebank.arXiv:2109.01982v3 [cs.CL] 29 Nov 2022Published as a conference paper at ICLR 2022 stack RNNs (Grefenstette et al., 2015;Joulin & Mikolov, 2015)which model deterministic stacks, being designed to learn one correct stack operation at each time step. One reason nondeterminism is important is that deterministic CFLs are a proper subset of CFLs. If the analogy with PDAs holds true, then equipping an RNN with a deterministic stack would only enable it to model deterministic CFLs, whereas a nondeterministic stack should enable it to model all CFLs. This is important for natural language processing, as human language is known to be high in syntactic ambiguity.Another benefit of nondeterminism, even on deterministic CFLs, applies to training. In order for a model to receive a reward for an action, it must try the action (that is, give it nonzero probability so that it receives gradient during backpropagation). For example, in the digit-recognition task, a classifier tries all ten digits, and is rewarded for the correct one. But in a stack-augmented model, the space of possible action sequences is very large. Whereas a deterministic stack can only try one of them, a nondeterministic stack can try all of them and always receives a reward for the correct one. But as explained in §3.1, because the NS-RNN's probability for an action sequence is the product of many probabilities, it can be extremely small, so the NS-RNN sometimes learns very slowly.ModelAgr.Lic. GPE GSE CE LDD LSTM, 256 units 0.667 0.446 0.330 0.397 0.482 0.414 LSTM, 258 units 0.658 0.447 0.335 0.375 0.518 0.357 LSTM, 267 units 0.667 0.497 0.343 0.446 0.411 0.350 JM (push hidden state) 0.640 0.408 0.296 0.310 0.464 0.352 JM (push learned) 0.684 0.439 0.340 0.408 0.482 0.395 NS, | | = 1, |Γ| = 2 0.588 0.452 0.298 0.391 0.339 0.418 NS, | | = 1, |Γ| = 3 0.623 0.467 0.400 0.413 0.393 0.354 NS, | | = 1, |Γ| = 4 0.640 0.497 0.331 0.375 0.571 0.340 NS, | | = 1, |Γ| = 5 0.605 0.514 0.394 0.413 0.589 0.344 NS, | | = 1, |Γ| = 6 0.632 0.424 0.408 0.391 0.464 0.399 NS, | | = 1, |Γ| = 7 0.719 0.470 0.351 0.473 0.500 0.344 NS, | | = 1, |Γ| = 11 0.640 0.432 0.329 0.424 0.500 0.413 NS, | | = 2, |Γ| = 2 0.702 0.388 0.329 0.446 0.446 0.371 NS, | | = 2, |Γ| = 3 0.658 0.527 0.367 0.446 0.518 0.411 NS, | | = 2, |Γ| = 4 0.632 0.464 0.345 0.386 0.518 0.387 NS, | | = 2, |Γ| = 5 0.711 0.464 0.307 0.413 0.518 0.355 NS, | | = 3, |Γ| = 2 0.711 0.528 0.349 0.435 0.518 0.406 NS, | | = 3, |Γ| = 3 0.746 0.439 0.316 0.375 0.411 0.376 NS, | | = 3, |Γ| = 4 0.702 0.450 0.364 0.484 0.536 0.369 RNS, | | = 1, |Γ| = 2 0.702 0.460 0.280 0.451 0.464 0.404 RNS, | | = 1, |Γ| = 3 0.649 0.427 0.438 0.418 0.446 0.347 RNS, | | = 1, |Γ| = 4 0.658 0.412 0.342 0.565 0.339 0.418 RNS, | | = 1, |Γ| = 5 0.728 0.449 0.370 0.429 0.482 0.371 RNS, | | = 1, |Γ| = 6 0.614 0.422 0.314 0.435 0.518 0.377 RNS, | | = 1, |Γ| = 7 0.649 0.460 0.374 0.337 0.411 0.404 RNS, | | = 1, |Γ| = 11 0.614 0.447 0.291 0.266 0.446 0.338 RNS, | | = 2, |Γ| = 2 0.649 0.417 0.365 0.375 0.339 0.334 RNS, | | = 2, |Γ| = 3 0.640 0.474 0.411 0.446 0.554 0.408 RNS, | | = 2, |Γ| = 4 0.658 0.469 0.336 0.326 0.500 0.403 RNS, | | = 2, |Γ| = 5 0.693 0.420 0.339 0.370 0.607 0.376 RNS, | | = 3, |Γ| = 2 0.579 0.435 0.295 0.440 0.554 0.445 RNS, | | = 3, |Γ| = 3 0.632 0.444 0.356 0.418 0.482 0.403 RNS, | | = 3, |Γ| = 4 0.588 0.427 0.342 0.353 0.482 0.373 | |
d17826787 | The ability to train large-scale neural networks has resulted in state-of-the-art performance in many areas of computer vision. These results have largely come from computational break throughs of two forms: model parallelism, e.g. GPU accelerated training, which has seen quick adoption in computer vision circles, and data parallelism, e.g. A-SGD, whose large scale has been used mostly in industry. We report early experiments with a system that makes use of both model parallelism and data parallelism, we call GPU A-SGD. We show using GPU A-SGD it is possible to speed up training of large convolutional neural networks useful for computer vision. We believe GPU A-SGD will make it possible to train larger networks on larger training sets in a reasonable amount of time. | GPU Asynchronous Stochastic Gradient Descent to Speed Up Neural Network Training |
d252683719 | Consider the problem of estimating the causal effect of some attribute of a text document; for example: what effect does writing a polite vs. rude email have on response time?To estimate a causal effect from observational data, we need to adjust for confounding aspects of the text that affect both the treatment and outcome-e.g., the topic or writing level of the text.These confounding aspects are unknown a priori, so it seems natural to adjust for the entirety of the text (e.g., using a transformer).However, causal identification and estimation procedures rely on the assumption of overlap: for all levels of the adjustment variables, there is randomness leftover so that every unit could have (not) received treatment.Since the treatment here is itself an attribute of the text, it is perfectly determined, and overlap is apparently violated.The purpose of this paper is to show how to handle causal identification and obtain robust causal estimation in the presence of apparent overlap violations.In brief, the idea is to use supervised representation learning to produce a data representation that preserves confounding information while eliminating information that is only predictive of the treatment.This representation then suffices for adjustment and satisfies overlap.Adapting results on non-parametric estimation, we find that this procedure is robust to conditional outcome misestimation, yielding a low-absolute-bias estimator with valid uncertainty quantification under weak conditions.Empirical results show strong improvements in bias and uncertainty quantification relative to the natural baseline.Code, demo data and a tutorial are available at https://github.com/gl-ybnbxb/TI-estimator. | CAUSAL ESTIMATION FOR TEXT DATA WITH (APPAR-ENT) OVERLAP VIOLATIONS |
d257365448 | Recent advancements in explainable machine learning provide effective and faithful solutions for interpreting model behaviors. However, many explanation methods encounter efficiency issues, which largely limit their deployments in practical scenarios. Real-time explainer (RTX) frameworks have thus been proposed to accelerate the model explanation process by learning a one-feed-forward explainer. Existing RTX frameworks typically build the explainer under the supervised learning paradigm, which requires large amounts of explanation labels as the ground truth. Considering that accurate explanation labels are usually hard to obtain due to constrained computational resources and limited human efforts, effective explainer training is still challenging in practice. In this work, we propose a COntrastive Real-Time eXplanation (CoRTX) framework to learn the explanation-oriented representation and relieve the intensive dependence of explainer training on explanation labels. Specifically, we design a synthetic strategy to select positive and negative instances for the learning of explanation. Theoretical analysis show that our selection strategy can benefit the contrastive learning process on explanation tasks. Experimental results on three real-world datasets further demonstrate the efficiency and efficacy of our proposed CoRTX framework. Our source code is available at: https://github.com/ynchuang/CoRTX-720 * These authors contributed equally to this work. | Published as a conference paper at ICLR 2023 CORTX: CONTRASTIVE FRAMEWORK FOR REAL- TIME EXPLANATION |
d255522680 | Various saliency map methods have been proposed to interpret and explain predictions of deep learning models. Saliency maps allow us to interpret which parts of the input signals have a strong influence on the prediction results. However, since a saliency map is obtained by complex computations in deep learning models, it is often difficult to know how reliable the saliency map itself is. In this study, we propose a method to quantify the reliability of a salient region in the form of p-values. Our idea is to consider a salient region as a selected hypothesis by the trained deep learning model and employ the selective inference framework. The proposed method can provably control the probability of false positive detections of salient regions. We demonstrate the validity of the proposed method through numerical examples in synthetic and real datasets.Furthermore, we develop a Keras-based framework for conducting the proposed selective inference for a wide class of CNNs without additional implementation cost. | Valid P -Value for Deep Learning-Driven Salient Region |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.