id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
2502.06719
Gaussian Approximation and Multiplier Bootstrap for Stochastic Gradient Descent
stat.ML cs.LG math.OC math.PR math.ST stat.TH
In this paper, we establish non-asymptotic convergence rates in the central limit theorem for Polyak-Ruppert-averaged iterates of stochastic gradient descent (SGD). Our analysis builds on the result of the Gaussian approximation for nonlinear statistics of independent random variables of Shao and Zhang (2022). Using this result, we prove the non-asymptotic validity of the multiplier bootstrap for constructing the confidence sets for the optimal solution of an optimization problem. In particular, our approach avoids the need to approximate the limiting covariance of Polyak-Ruppert SGD iterates, which allows us to derive approximation rates in convex distance of order up to $1/\sqrt{n}$.
2502.06722
HetSwarm: Cooperative Navigation of Heterogeneous Swarm in Dynamic and Dense Environments through Impedance-based Guidance
cs.RO
With the growing demand for efficient logistics and warehouse management, unmanned aerial vehicles (UAVs) are emerging as a valuable complement to automated guided vehicles (AGVs). UAVs enhance efficiency by navigating dense environments and operating at varying altitudes. However, their limited flight time, battery life, and payload capacity necessitate a supporting ground station. To address these challenges, we propose HetSwarm, a heterogeneous multi-robot system that combines a UAV and a mobile ground robot for collaborative navigation in cluttered and dynamic conditions. Our approach employs an artificial potential field (APF)-based path planner for the UAV, allowing it to dynamically adjust its trajectory in real time. The ground robot follows this path while maintaining connectivity through impedance links, ensuring stable coordination. Additionally, the ground robot establishes temporal impedance links with low-height ground obstacles to avoid local collisions, as these obstacles do not interfere with the UAV's flight. Experimental validation of HetSwarm in diverse environmental conditions demonstrated a 90% success rate across 30 test cases. The ground robot exhibited an average deviation of 45 cm near obstacles, confirming effective collision avoidance. Extensive simulations in the Gym PyBullet environment further validated the robustness of our system for real-world applications, demonstrating its potential for dynamic, real-time task execution in cluttered environments.
2502.06725
AgilePilot: DRL-Based Drone Agent for Real-Time Motion Planning in Dynamic Environments by Leveraging Object Detection
cs.RO
Autonomous drone navigation in dynamic environments remains a critical challenge, especially when dealing with unpredictable scenarios including fast-moving objects with rapidly changing goal positions. While traditional planners and classical optimisation methods have been extensively used to address this dynamic problem, they often face real-time, unpredictable changes that ultimately leads to sub-optimal performance in terms of adaptiveness and real-time decision making. In this work, we propose a novel motion planner, AgilePilot, based on Deep Reinforcement Learning (DRL) that is trained in dynamic conditions, coupled with real-time Computer Vision (CV) for object detections during flight. The training-to-deployment framework bridges the Sim2Real gap, leveraging sophisticated reward structures that promotes both safety and agility depending upon environment conditions. The system can rapidly adapt to changing environments, while achieving a maximum speed of 3.0 m/s in real-world scenarios. In comparison, our approach outperforms classical algorithms such as Artificial Potential Field (APF) based motion planner by 3 times, both in performance and tracking accuracy of dynamic targets by using velocity predictions while exhibiting 90% success rate in 75 conducted experiments. This work highlights the effectiveness of DRL in tackling real-time dynamic navigation challenges, offering intelligent safety and agility.
2502.06726
Rough Stochastic Pontryagin Maximum Principle and an Indirect Shooting Method
math.OC cs.RO cs.SY eess.SY math.PR
We derive first-order Pontryagin optimality conditions for stochastic optimal control with deterministic controls for systems modeled by rough differential equations (RDE) driven by Gaussian rough paths. This Pontryagin Maximum Principle (PMP) applies to systems following stochastic differential equations (SDE) driven by Brownian motion, yet it does not rely on forward-backward SDEs and involves the same Hamiltonian as the deterministic PMP. The proof consists of first deriving various integrable error bounds for solutions to nonlinear and linear RDEs by leveraging recent results on Gaussian rough paths. The PMP then follows using standard techniques based on needle-like variations. As an application, we propose the first indirect shooting method for nonlinear stochastic optimal control and show that it converges 10x faster than a direct method on a stabilization task.
2502.06727
Application of Artificial Intelligence (AI) in Civil Engineering
cs.AI
Hard computing generally deals with precise data, which provides ideal solutions to problems. However, in the civil engineering field, amongst other disciplines, that is not always the case as real-world systems are continuously changing. Here lies the need to explore soft computing methods and artificial intelligence to solve civil engineering shortcomings. The integration of advanced computational models, including Artificial Neural Networks (ANNs), Fuzzy Logic, Genetic Algorithms (GAs), and Probabilistic Reasoning, has revolutionized the domain of civil engineering. These models have significantly advanced diverse sub-fields by offering innovative solutions and improved analysis capabilities. Sub-fields such as: slope stability analysis, bearing capacity, water quality and treatment, transportation systems, air quality, structural materials, etc. ANNs predict non-linearities and provide accurate estimates. Fuzzy logic uses an efficient decision-making process to provide a more precise assessment of systems. Lastly, while GAs optimizes models (based on evolutionary processes) for better outcomes, probabilistic reasoning lowers their statistical uncertainties.
2502.06728
FlexDeMo: Decoupled Momentum Optimization for Fully and Hybrid Sharded Training
cs.LG cs.AI
Training large neural network models requires extensive computational resources, often distributed across several nodes and accelerators. Recent findings suggest that it may be sufficient to only exchange the fast moving components of the gradients, while accumulating momentum locally (Decoupled Momentum, or DeMo). However, when considering larger models that do not fit on a single accelerate, the exchange of gradient information and the integration of DeMo needs to be reconsidered. Here, we propose employing a hybrid strategy, FlexDeMo, whereby nodes fully synchronize locally between different GPUs and inter-node communication is improved through only using the fast-moving components. This effectively combines previous hybrid sharding strategies with the advantages of decoupled momentum. Our experimental results show that FlexDeMo is on par with AdamW in terms of validation loss, demonstrating its viability.
2502.06733
Dynamic Loss-Based Sample Reweighting for Improved Large Language Model Pretraining
cs.LG cs.AI
Pretraining large language models (LLMs) on vast and heterogeneous datasets is crucial for achieving state-of-the-art performance across diverse downstream tasks. However, current training paradigms treat all samples equally, overlooking the importance or relevance of individual samples throughout the training process. Existing reweighting strategies, which primarily focus on group-level data importance, fail to leverage fine-grained instance-level information and do not adapt dynamically to individual sample importance as training progresses. In this paper, we introduce novel algorithms for dynamic, instance-level data reweighting aimed at improving both the efficiency and effectiveness of LLM pretraining. Our methods adjust the weight of each training sample based on its loss value in an online fashion, allowing the model to dynamically focus on more informative or important samples at the current training stage. In particular, our framework allows us to systematically devise reweighting strategies deprioritizing redundant or uninformative data, which we find tend to work best. Furthermore, we develop a new theoretical framework for analyzing the impact of loss-based reweighting on the convergence of gradient-based optimization, providing the first formal characterization of how these strategies affect convergence bounds. We empirically validate our approach across a spectrum of tasks, from pretraining 7B and 1.4B parameter LLMs to smaller-scale language models and linear regression problems, demonstrating that our loss-based reweighting approach can lead to faster convergence and significantly improved performance.
2502.06734
Se\~norita-2M: A High-Quality Instruction-based Dataset for General Video Editing by Video Specialists
cs.CV
Recent advancements in video generation have spurred the development of video editing techniques, which can be divided into inversion-based and end-to-end methods. However, current video editing methods still suffer from several challenges. Inversion-based methods, though training-free and flexible, are time-consuming during inference, struggle with fine-grained editing instructions, and produce artifacts and jitter. On the other hand, end-to-end methods, which rely on edited video pairs for training, offer faster inference speeds but often produce poor editing results due to a lack of high-quality training video pairs. In this paper, to close the gap in end-to-end methods, we introduce Se\~norita-2M, a high-quality video editing dataset. Se\~norita-2M consists of approximately 2 millions of video editing pairs. It is built by crafting four high-quality, specialized video editing models, each crafted and trained by our team to achieve state-of-the-art editing results. We also propose a filtering pipeline to eliminate poorly edited video pairs. Furthermore, we explore common video editing architectures to identify the most effective structure based on current pre-trained generative model. Extensive experiments show that our dataset can help to yield remarkably high-quality video editing results. More details are available at https://senorita.github.io.
2502.06735
Enhancing Pneumonia Diagnosis and Severity Assessment through Deep Learning: A Comprehensive Approach Integrating CNN Classification and Infection Segmentation
cs.CV
Lung disease poses a substantial global health challenge, with pneumonia being a prevalent concern. This research focuses on leveraging deep learning techniques to detect and assess pneumonia, addressing two interconnected objectives. Initially, Convolutional Neural Network (CNN) models are introduced for pneumonia classification, emphasizing the necessity of comprehensive diagnostic assessments considering COVID-19. Subsequently, the study advocates for the utilization of deep learning-based segmentation to determine the severity of infection. This dual-pronged approach offers valuable insights for medical professionals, facilitating a more nuanced understanding and effective treatment of pneumonia. Integrating deep learning aims to elevate the accuracy and efficiency of pneumonia detection, thereby contributing to enhanced healthcare outcomes on a global scale.
2502.06736
Low-power Spike-based Wearable Analytics on RRAM Crossbars
cs.ET cs.AI cs.AR
This work introduces a spike-based wearable analytics system utilizing Spiking Neural Networks (SNNs) deployed on an In-memory Computing engine based on RRAM crossbars, which are known for their compactness and energy-efficiency. Given the hardware constraints and noise characteristics of the underlying RRAM crossbars, we propose online adaptation of pre-trained SNNs in real-time using Direct Feedback Alignment (DFA) against traditional backpropagation (BP). Direct Feedback Alignment (DFA) learning, that allows layer-parallel gradient computations, acts as a fast, energy & area-efficient method for online adaptation of SNNs on RRAM crossbars, unleashing better algorithmic performance against those adapted using BP. Through extensive simulations using our in-house hardware evaluation engine called DFA_Sim, we find that DFA achieves upto 64.1% lower energy consumption, 10.1% lower area overhead, and a 2.1x reduction in latency compared to BP, while delivering upto 7.55% higher inference accuracy on human activity recognition (HAR) tasks.
2502.06737
VersaPRM: Multi-Domain Process Reward Model via Synthetic Reasoning Data
cs.LG
Process Reward Models (PRMs) have proven effective at enhancing mathematical reasoning for Large Language Models (LLMs) by leveraging increased inference-time computation. However, they are predominantly trained on mathematical data and their generalizability to non-mathematical domains has not been rigorously studied. In response, this work first shows that current PRMs have poor performance in other domains. To address this limitation, we introduce VersaPRM, a multi-domain PRM trained on synthetic reasoning data generated using our novel data generation and annotation method. VersaPRM achieves consistent performance gains across diverse domains. For instance, in the MMLU-Pro category of Law, VersaPRM via weighted majority voting, achieves a 7.9% performance gain over the majority voting baseline -- surpassing Qwen2.5-Math-PRM's gain of 1.3%. We further contribute to the community by open-sourcing all data, code and models for VersaPRM.
2502.06738
Resurrecting saturated LLM benchmarks with adversarial encoding
cs.LG
Recent work showed that small changes in benchmark questions can reduce LLMs' reasoning and recall. We explore two such changes: pairing questions and adding more answer options, on three benchmarks: WMDP-bio, GPQA, and MMLU variants. We find that for more capable models, these predictably reduce performance, essentially heightening the performance ceiling of a benchmark and unsaturating it again. We suggest this approach can resurrect old benchmarks.
2502.06739
A note on the physical interpretation of neural PDE's
cs.LG cond-mat.dis-nn physics.comp-ph
We highlight a formal and substantial analogy between Machine Learning (ML) algorithms and discrete dynamical systems (DDS) in relaxation form. The analogy offers a transparent interpretation of the weights in terms of physical information-propagation processes and identifies the model function of the forward ML step with the local attractor of the corresponding discrete dynamics. Besides improving the explainability of current ML applications, this analogy may also facilitate the development of a new class ML algorithms with a reduced number of weights.
2502.06741
ViSIR: Vision Transformer Single Image Reconstruction Method for Earth System Models
cs.CV
Purpose: Earth system models (ESMs) integrate the interactions of the atmosphere, ocean, land, ice, and biosphere to estimate the state of regional and global climate under a wide variety of conditions. The ESMs are highly complex, and thus, deep neural network architectures are used to model the complexity and store the down-sampled data. In this paper, we propose the Vision Transformer Sinusoidal Representation Networks (ViSIR) to improve the single image SR (SR) reconstruction task for the ESM data. Methods: ViSIR combines the SR capability of Vision Transformers (ViT) with the high-frequency detail preservation of the Sinusoidal Representation Network (SIREN) to address the spectral bias observed in SR tasks. Results: The ViSIR outperforms ViT by 4.1 dB, SIREN by 7.5 dB, and SR-Generative Adversarial (SR-GANs) by 7.1dB PSNR on average for three different measurements. Conclusion: The proposed ViSIR is evaluated and compared with state-of-the-art methods. The results show that the proposed algorithm is outperforming other methods in terms of Mean Square Error(MSE), Peak-Signal-to-Noise-Ratio(PSNR), and Structural Similarity Index Measure(SSIM).
2502.06742
Gradient Multi-Normalization for Stateless and Scalable LLM Training
cs.LG cs.AI
Training large language models (LLMs) typically relies on adaptive optimizers like Adam (Kingma & Ba, 2015) which store additional state information to accelerate convergence but incur significant memory overhead. Recent efforts, such as SWAN (Ma et al., 2024) address this by eliminating the need for optimizer states while achieving performance comparable to Adam via a multi-step preprocessing procedure applied to instantaneous gradients. Motivated by the success of SWAN, we introduce a novel framework for designing stateless optimizers that normalizes stochastic gradients according to multiple norms. To achieve this, we propose a simple alternating scheme to enforce the normalization of gradients w.r.t these norms. We show that our procedure can produce, up to an arbitrary precision, a fixed-point of the problem, and that SWAN is a particular instance of our approach with carefully chosen norms, providing a deeper understanding of its design. However, SWAN's computationally expensive whitening/orthogonalization step limit its practicality for large LMs. Using our principled perspective, we develop of a more efficient, scalable, and practical stateless optimizer. Our algorithm relaxes the properties of SWAN, significantly reducing its computational cost while retaining its memory efficiency, making it applicable to training large-scale models. Experiments on pre-training LLaMA models with up to 1 billion parameters demonstrate a 3X speedup over Adam with significantly reduced memory requirements, outperforming other memory-efficient baselines.
2502.06747
Wandering around: A bioinspired approach to visual attention through object motion sensitivity
cs.CV
Active vision enables dynamic visual perception, offering an alternative to static feedforward architectures in computer vision, which rely on large datasets and high computational resources. Biological selective attention mechanisms allow agents to focus on salient Regions of Interest (ROIs), reducing computational demand while maintaining real-time responsiveness. Event-based cameras, inspired by the mammalian retina, enhance this capability by capturing asynchronous scene changes enabling efficient low-latency processing. To distinguish moving objects while the event-based camera is in motion the agent requires an object motion segmentation mechanism to accurately detect targets and center them in the visual field (fovea). Integrating event-based sensors with neuromorphic algorithms represents a paradigm shift, using Spiking Neural Networks to parallelize computation and adapt to dynamic environments. This work presents a Spiking Convolutional Neural Network bioinspired attention system for selective attention through object motion sensitivity. The system generates events via fixational eye movements using a Dynamic Vision Sensor integrated into the Speck neuromorphic hardware, mounted on a Pan-Tilt unit, to identify the ROI and saccade toward it. The system, characterized using ideal gratings and benchmarked against the Event Camera Motion Segmentation Dataset, reaches a mean IoU of 82.2% and a mean SSIM of 96% in multi-object motion segmentation. The detection of salient objects reaches 88.8% accuracy in office scenarios and 89.8% in low-light conditions on the Event-Assisted Low-Light Video Object Segmentation Dataset. A real-time demonstrator shows the system's 0.12 s response to dynamic scenes. Its learning-free design ensures robustness across perceptual scenes, making it a reliable foundation for real-time robotic applications serving as a basis for more complex architectures.
2502.06748
Institutional Preferences in the Laboratory
cs.SI cs.GT
Getting a group to adopt cooperative norms is an enduring challenge. But in real-world settings, individuals don't just passively accept static environments, they act both within and upon the social systems that structure their interactions. Should we expect the dynamism of player-driven changes to the "rules of the game" to hinder cooperation -- because of the substantial added complexity -- or help it, as prosocial agents tweak their environment toward non-zero-sum games? We introduce a laboratory setting to test whether groups can guide themselves to cooperative outcomes by manipulating the environmental parameters that shape their emergent cooperation process. We test for cooperation in a set of economic games that impose different social dilemmas. These games vary independently in the institutional features of stability, efficiency, and fairness. By offering agency over behavior along with second-order agency over the rules of the game, we understand emergent cooperation in naturalistic settings in which the rules of the game are themselves dynamic and subject to choice. The literature on transfer learning in games suggests that interactions between features are important and might aid or hinder the transfer of cooperative learning to new settings.
2502.06749
Incentivizing Desirable Effort Profiles in Strategic Classification: The Role of Causality and Uncertainty
cs.GT cs.CY cs.LG
We study strategic classification in binary decision-making settings where agents can modify their features in order to improve their classification outcomes. Importantly, our work considers the causal structure across different features, acknowledging that effort in a given feature may affect other features. The main goal of our work is to understand \emph{when and how much agent effort is invested towards desirable features}, and how this is influenced by the deployed classifier, the causal structure of the agent's features, their ability to modify them, and the information available to the agent about the classifier and the feature causal graph. In the complete information case, when agents know the classifier and the causal structure of the problem, we derive conditions ensuring that rational agents focus on features favored by the principal. We show that designing classifiers to induce desirable behavior is generally non-convex, though tractable in special cases. We also extend our analysis to settings where agents have incomplete information about the classifier or the causal graph. While optimal effort selection is again a non-convex problem under general uncertainty, we highlight special cases of partial uncertainty where this selection problem becomes tractable. Our results indicate that uncertainty drives agents to favor features with higher expected importance and lower variance, potentially misaligning with principal preferences. Finally, numerical experiments based on a cardiovascular disease risk study illustrate how to incentivize desirable modifications under uncertainty.
2502.06750
Accelerating Data Processing and Benchmarking of AI Models for Pathology
cs.CV
Advances in foundation modeling have reshaped computational pathology. However, the increasing number of available models and lack of standardized benchmarks make it increasingly complex to assess their strengths, limitations, and potential for further development. To address these challenges, we introduce a new suite of software tools for whole-slide image processing, foundation model benchmarking, and curated publicly available tasks. We anticipate that these resources will promote transparency, reproducibility, and continued progress in the field.
2502.06751
What makes a good feedforward computational graph?
cs.LG cs.AI cs.SI stat.ML
As implied by the plethora of literature on graph rewiring, the choice of computational graph employed by a neural network can make a significant impact on its downstream performance. Certain effects related to the computational graph, such as under-reaching and over-squashing, may even render the model incapable of learning certain functions. Most of these effects have only been thoroughly studied in the domain of undirected graphs; however, recent years have seen a significant rise in interest in feedforward computational graphs: directed graphs without any back edges. In this paper, we study the desirable properties of a feedforward computational graph, discovering two important complementary measures: fidelity and mixing time, and evaluating a few popular choices of graphs through the lens of these measures. Our study is backed by both theoretical analyses of the metrics' asymptotic behaviour for various graphs, as well as correlating these metrics to the performance of trained neural network models using the corresponding graphs.
2502.06753
Case for a unified surrogate modelling framework in the age of AI
stat.CO cs.LG
Surrogate models are widely used in natural sciences, engineering, and machine learning to approximate complex systems and reduce computational costs. However, the current landscape lacks standardisation across key stages of the pipeline, including data collection, sampling design, model class selection, evaluation metrics, and downstream task performance analysis. This fragmentation limits reproducibility, reliability, and cross-domain applicability. The issue has only been exacerbated by the AI revolution and a new suite of surrogate model classes that it offers. In this position paper, we argue for the urgent need for a unified framework to guide the development and evaluation of surrogate models. We outline essential steps for constructing a comprehensive pipeline and discuss alternative perspectives, such as the benefits of domain-specific frameworks. By advocating for a standardised approach, this paper seeks to improve the reliability of surrogate modelling, foster cross-disciplinary knowledge transfer, and, as a result, accelerate scientific progress.
2502.06755
Sparse Autoencoders for Scientifically Rigorous Interpretation of Vision Models
cs.CV
To truly understand vision models, we must not only interpret their learned features but also validate these interpretations through controlled experiments. Current approaches either provide interpretable features without the ability to test their causal influence, or enable model editing without interpretable controls. We present a unified framework using sparse autoencoders (SAEs) that bridges this gap, allowing us to discover human-interpretable visual features and precisely manipulate them to test hypotheses about model behavior. By applying our method to state-of-the-art vision models, we reveal key differences in the semantic abstractions learned by models with different pre-training objectives. We then demonstrate the practical usage of our framework through controlled interventions across multiple vision tasks. We show that SAEs can reliably identify and manipulate interpretable visual features without model re-training, providing a powerful tool for understanding and controlling vision model behavior. We provide code, demos and models on our project website: https://osu-nlp-group.github.io/SAE-V.
2502.06756
SAMRefiner: Taming Segment Anything Model for Universal Mask Refinement
cs.CV
In this paper, we explore a principal way to enhance the quality of widely pre-existing coarse masks, enabling them to serve as reliable training data for segmentation models to reduce the annotation cost. In contrast to prior refinement techniques that are tailored to specific models or tasks in a close-world manner, we propose SAMRefiner, a universal and efficient approach by adapting SAM to the mask refinement task. The core technique of our model is the noise-tolerant prompting scheme. Specifically, we introduce a multi-prompt excavation strategy to mine diverse input prompts for SAM (i.e., distance-guided points, context-aware elastic bounding boxes, and Gaussian-style masks) from initial coarse masks. These prompts can collaborate with each other to mitigate the effect of defects in coarse masks. In particular, considering the difficulty of SAM to handle the multi-object case in semantic segmentation, we introduce a split-then-merge (STM) pipeline. Additionally, we extend our method to SAMRefiner++ by introducing an additional IoU adaption step to further boost the performance of the generic SAMRefiner on the target dataset. This step is self-boosted and requires no additional annotation. The proposed framework is versatile and can flexibly cooperate with existing segmentation methods. We evaluate our mask framework on a wide range of benchmarks under different settings, demonstrating better accuracy and efficiency. SAMRefiner holds significant potential to expedite the evolution of refinement tools. Our code is available at https://github.com/linyq2117/SAMRefiner.
2502.06759
Rationalization Models for Text-to-SQL
cs.CL cs.AI cs.DB
We introduce a framework for generating Chain-of-Thought (CoT) rationales to enhance text-to-SQL model fine-tuning. These rationales consist of intermediate SQL statements and explanations, serving as incremental steps toward constructing the final SQL query. The process begins with manually annotating a small set of examples, which are then used to prompt a large language model in an iterative, dynamic few-shot knowledge distillation procedure from a teacher model. A rationalization model is subsequently trained on the validated decomposed queries, enabling extensive synthetic CoT annotations for text-to-SQL datasets. To evaluate the approach, we fine-tune small language models with and without these rationales on the BIRD dataset. Results indicate that step-by-step query generation improves execution accuracy, especially for moderately and highly complex queries, while also enhancing explainability.
2502.06760
Infinite-Horizon Value Function Approximation for Model Predictive Control
cs.RO
Model Predictive Control has emerged as a popular tool for robots to generate complex motions. However, the real-time requirement has limited the use of hard constraints and large preview horizons, which are necessary to ensure safety and stability. In practice, practitioners have to carefully design cost functions that can imitate an infinite horizon formulation, which is tedious and often results in local minima. In this work, we study how to approximate the infinite horizon value function of constrained optimal control problems with neural networks using value iteration and trajectory optimization. Furthermore, we demonstrate how using this value function approximation as a terminal cost provides global stability to the model predictive controller. The approach is validated on two toy problems and a real-world scenario with online obstacle avoidance on an industrial manipulator where the value function is conditioned to the goal and obstacle.
2502.06761
When, Where and Why to Average Weights?
cs.LG
Averaging checkpoints along the training trajectory is a simple yet powerful approach to improve the generalization performance of Machine Learning models and reduce training time. Motivated by these potential gains, and in an effort to fairly and thoroughly benchmark this technique, we present an extensive evaluation of averaging techniques in modern Deep Learning, which we perform using AlgoPerf \citep{dahl_benchmarking_2023}, a large-scale benchmark for optimization algorithms. We investigate whether weight averaging can reduce training time, improve generalization, and replace learning rate decay, as suggested by recent literature. Our evaluation across seven architectures and datasets reveals that averaging significantly accelerates training and yields considerable efficiency gains, at the price of a minimal implementation and memory cost, while mildly improving generalization across all considered workloads. Finally, we explore the relationship between averaging and learning rate annealing and show how to optimally combine the two to achieve the best performances.
2502.06764
History-Guided Video Diffusion
cs.LG cs.CV
Classifier-free guidance (CFG) is a key technique for improving conditional generation in diffusion models, enabling more accurate control while enhancing sample quality. It is natural to extend this technique to video diffusion, which generates video conditioned on a variable number of context frames, collectively referred to as history. However, we find two key challenges to guiding with variable-length history: architectures that only support fixed-size conditioning, and the empirical observation that CFG-style history dropout performs poorly. To address this, we propose the Diffusion Forcing Transformer (DFoT), a video diffusion architecture and theoretically grounded training objective that jointly enable conditioning on a flexible number of history frames. We then introduce History Guidance, a family of guidance methods uniquely enabled by DFoT. We show that its simplest form, vanilla history guidance, already significantly improves video generation quality and temporal consistency. A more advanced method, history guidance across time and frequency further enhances motion dynamics, enables compositional generalization to out-of-distribution history, and can stably roll out extremely long videos. Website: https://boyuan.space/history-guidance
2502.06765
Are all models wrong? Fundamental limits in distribution-free empirical model falsification
math.ST cs.LG stat.ML stat.TH
In statistics and machine learning, when we train a fitted model on available data, we typically want to ensure that we are searching within a model class that contains at least one accurate model -- that is, we would like to ensure an upper bound on the model class risk (the lowest possible risk that can be attained by any model in the class). However, it is also of interest to establish lower bounds on the model class risk, for instance so that we can determine whether our fitted model is at least approximately optimal within the class, or, so that we can decide whether the model class is unsuitable for the particular task at hand. Particularly in the setting of interpolation learning where machine learning models are trained to reach zero error on the training data, we might ask if, at the very least, a positive lower bound on the model class risk is possible -- or are we unable to detect that "all models are wrong"? In this work, we answer these questions in a distribution-free setting by establishing a model-agnostic, fundamental hardness result for the problem of constructing a lower bound on the best test error achievable over a model class, and examine its implications on specific model classes such as tree-based methods and linear regression.
2502.06766
Exploiting Sparsity for Long Context Inference: Million Token Contexts on Commodity GPUs
cs.CL
There is growing demand for performing inference with hundreds of thousands of input tokens on trained transformer models. Inference at this extreme scale demands significant computational resources, hindering the application of transformers at long contexts on commodity (i.e not data center scale) hardware. To address the inference time costs associated with running self-attention based transformer language models on long contexts and enable their adoption on widely available hardware, we propose a tunable mechanism that reduces the cost of the forward pass by attending to only the most relevant tokens at every generation step using a top-k selection mechanism. We showcase the efficiency gains afforded by our method by performing inference on context windows up to 1M tokens using approximately 16GB of GPU RAM. Our experiments reveal that models are capable of handling the sparsity induced by the reduced number of keys and values. By attending to less than 2% of input tokens, we achieve over 95% of model performance on common benchmarks (RULER, AlpacaEval, and Open LLM Leaderboard).
2502.06768
Train for the Worst, Plan for the Best: Understanding Token Ordering in Masked Diffusions
cs.LG
In recent years, masked diffusion models (MDMs) have emerged as a promising alternative approach for generative modeling over discrete domains. Compared to autoregressive models (ARMs), MDMs trade off complexity at training time with flexibility at inference time. At training time, they must learn to solve an exponentially large number of infilling problems, but at inference time, they can decode tokens in essentially arbitrary order. In this work, we closely examine these two competing effects. On the training front, we theoretically and empirically demonstrate that MDMs indeed train on computationally intractable subproblems compared to their autoregressive counterparts. On the inference front, we show that a suitable strategy for adaptively choosing the token decoding order significantly enhances the capabilities of MDMs, allowing them to sidestep hard subproblems. On logic puzzles like Sudoku, we show that adaptive inference can boost solving accuracy in pretrained MDMs from $<7$% to $\approx 90$%, even outperforming ARMs with $7\times$ as many parameters and that were explicitly trained via teacher forcing to learn the right order of decoding.
2502.06770
Parameter-Dependent Control Lyapunov Functions for Stabilizing Nonlinear Parameter-Varying Systems
math.OC cs.SY eess.SY
This paper introduces the concept of parameter-dependent (PD) control Lyapunov functions (CLFs) for gainscheduled stabilization of nonlinear parameter-varying (NPV) systems under input constraints. It shows that given a PD-CLF, a locally Lipschitz control law can be constructed by solving a robust quadratic program. For polynomial control-affine NPV systems, it provides convex conditions, based on the sum of squares (SOS) programming, to jointly synthesize both a PD-CLF and a PD controller, aimed at maximizing the PD region of stabilization. Input constraints can be straightforwardly incorporated into the synthesis procedure. Unlike traditional linear parameter-varying (LPV) methods that rely on linearization or over-approximation to get an LPV model, the proposed framework fully captures the nonlinearities of the system dynamics. Simulation results validate the efficacy of the method, showcasing its potential for stabilizing NPV systems under input constraints.
2502.06771
Unsupervised Particle Tracking with Neuromorphic Computing
hep-ex cs.ET cs.LG cs.NE
We study the application of a neural network architecture for identifying charged particle trajectories via unsupervised learning of delays and synaptic weights using a spike-time-dependent plasticity rule. In the considered model, the neurons receive time-encoded information on the position of particle hits in a tracking detector for a particle collider, modeled according to the geometry of the Compact Muon Solenoid Phase II detector. We show how a spiking neural network is capable of successfully identifying in a completely unsupervised way the signal left by charged particles in the presence of conspicuous noise from accidental or combinatorial hits. These results open the way to applications of neuromorphic computing to particle tracking, motivating further studies into its potential for real-time, low-power particle tracking in future high-energy physics experiments.
2502.06772
ReasonFlux: Hierarchical LLM Reasoning via Scaling Thought Templates
cs.CL cs.AI cs.LG
We present that hierarchical LLM reasoning via scaling thought templates can effectively optimize the reasoning search space and outperform the mathematical reasoning capabilities of powerful LLMs like OpenAI o1-preview and DeepSeek V3. We train our ReasonFlux-32B model with only 8 GPUs and introduces three innovations: (i) a structured and generic thought template library, containing around 500 high-level thought templates capable of generalizing to similar or relevant reasoning problems; (ii) performing hierarchical reinforcement learning on a sequence of thought templates instead of long CoTs, optimizing a base LLM to plan out an optimal template trajectory for gradually handling complex problems; (iii) a brand new inference scaling system that enables hierarchical LLM reasoning by adaptively scaling thought templates at inference time. With a template trajectory containing sequential thought templates, our ReasonFlux-32B significantly advances math reasoning capabilities to state-of-the-art levels. Notably, on the MATH benchmark, it achieves an accuracy of 91.2% and surpasses o1-preview by 6.7%. On the USA Math Olympiad (AIME) benchmark, ReasonFlux-32B solves an average of 56.7% of problems, surpassing o1-preview and DeepSeek-V3 by 27% and 45%, respectively. Code: https://github.com/Gen-Verse/ReasonFlux
2502.06773
On the Emergence of Thinking in LLMs I: Searching for the Right Intuition
cs.AI cs.CL cs.LG
Recent AI advancements, such as OpenAI's new models, are transforming LLMs into LRMs (Large Reasoning Models) that perform reasoning during inference, taking extra time and compute for higher-quality outputs. We aim to uncover the algorithmic framework for training LRMs. Methods like self-consistency, PRM, and AlphaZero suggest reasoning as guided search. We ask: what is the simplest, most scalable way to enable search in LLMs? We propose a post-training framework called Reinforcement Learning via Self-Play (RLSP). RLSP involves three steps: (1) supervised fine-tuning with human or synthetic demonstrations of the reasoning process, (2) using an exploration reward signal to encourage diverse and efficient reasoning behaviors, and (3) RL training with an outcome verifier to ensure correctness while preventing reward hacking. Our key innovation is to decouple exploration and correctness signals during PPO training, carefully balancing them to improve performance and efficiency. Empirical studies in the math domain show that RLSP improves reasoning. On the Llama-3.1-8B-Instruct model, RLSP can boost performance by 23% in MATH-500 test set; On AIME 2024 math problems, Qwen2.5-32B-Instruct improved by 10% due to RLSP. However, a more important finding of this work is that the models trained using RLSP, even with the simplest exploration reward that encourages the model to take more intermediate steps, showed several emergent behaviors such as backtracking, exploration of ideas, and verification. These findings demonstrate that RLSP framework might be enough to enable emergence of complex reasoning abilities in LLMs when scaled. Lastly, we propose a theory as to why RLSP search strategy is more suitable for LLMs inspired by a remarkable result that says CoT provably increases computational power of LLMs, which grows as the number of steps in CoT \cite{li2024chain,merrill2023expresssive}.
2502.06774
ENFORCE: Exact Nonlinear Constrained Learning with Adaptive-depth Neural Projection
cs.LG
Ensuring neural networks adhere to domain-specific constraints is crucial for addressing safety and ethical concerns while also enhancing prediction accuracy. Despite the nonlinear nature of most real-world tasks, existing methods are predominantly limited to affine or convex constraints. We introduce ENFORCE, a neural network architecture that guarantees predictions to satisfy nonlinear constraints exactly. ENFORCE is trained with standard unconstrained gradient-based optimizers (e.g., Adam) and leverages autodifferentiation and local neural projections to enforce any $\mathcal{C}^1$ constraint to arbitrary tolerance $\epsilon$. We build an adaptive-depth neural projection (AdaNP) module that dynamically adjusts its complexity to suit the specific problem and the required tolerance levels. ENFORCE guarantees satisfaction of equality constraints that are nonlinear in both inputs and outputs of the neural network with minimal (and adjustable) computational cost.
2502.06775
Enhancing Performance of Explainable AI Models with Constrained Concept Refinement
cs.LG
The trade-off between accuracy and interpretability has long been a challenge in machine learning (ML). This tension is particularly significant for emerging interpretable-by-design methods, which aim to redesign ML algorithms for trustworthy interpretability but often sacrifice accuracy in the process. In this paper, we address this gap by investigating the impact of deviations in concept representations-an essential component of interpretable models-on prediction performance and propose a novel framework to mitigate these effects. The framework builds on the principle of optimizing concept embeddings under constraints that preserve interpretability. Using a generative model as a test-bed, we rigorously prove that our algorithm achieves zero loss while progressively enhancing the interpretability of the resulting model. Additionally, we evaluate the practical performance of our proposed framework in generating explainable predictions for image classification tasks across various benchmarks. Compared to existing explainable methods, our approach not only improves prediction accuracy while preserving model interpretability across various large-scale benchmarks but also achieves this with significantly lower computational cost.
2502.06776
Towards Internet-Scale Training For Agents
cs.LG cs.AI
The predominant approach for training web navigation agents gathers human demonstrations for a set of popular websites and hand-written tasks, but it is becoming clear that human data are an inefficient resource. We develop a pipeline to facilitate Internet-scale training for agents without laborious human annotations. In the first stage, an LLM generates tasks for 150k diverse websites. In the next stage, LLM agents complete tasks and produce trajectories. In the final stage, an LLM reviews the trajectories and judges their success. Language models are competitive with human annotators, detecting and filtering out harmful content with an accuracy of 97%, generating feasible tasks with an 89% rate, and judging successful trajectories with an 82.6% accuracy. Scaling the pipeline, agents based on Llama 3.1 70B solve 16.7% of tasks for 150k sites. Training on the data generated by our pipeline is competitive with training on human demonstrations. In data-limited settings derived from Mind2Web and WebLINX, we improve Step Accuracy by up to +89.5% and +122.1% respectively for agents trained on mixtures of data from our pipeline, and human data. When training agents with all available human data from these benchmarks, agents fail to generalize to diverse real sites, and adding our data improves their generalization by +149.0% for WebLINX and +156.3% for Mind2Web. Code will be available at: data-for-agents.github.io.
2502.06777
Learning an Optimal Assortment Policy under Observational Data
stat.ML cs.LG math.OC math.ST stat.TH
We study the fundamental problem of offline assortment optimization under the Multinomial Logit (MNL) model, where sellers must determine the optimal subset of the products to offer based solely on historical customer choice data. While most existing approaches to learning-based assortment optimization focus on the online learning of the optimal assortment through repeated interactions with customers, such exploration can be costly or even impractical in many real-world settings. In this paper, we consider the offline learning paradigm and investigate the minimal data requirements for efficient offline assortment optimization. To this end, we introduce Pessimistic Rank-Breaking (PRB), an algorithm that combines rank-breaking with pessimistic estimation. We prove that PRB is nearly minimax optimal by establishing the tight suboptimality upper bound and a nearly matching lower bound. This further shows that "optimal item coverage" - where each item in the optimal assortment appears sufficiently often in the historical data - is both sufficient and necessary for efficient offline learning. This significantly relaxes the previous requirement of observing the complete optimal assortment in the data. Our results provide fundamental insights into the data requirements for offline assortment optimization under the MNL model.
2502.06779
KARST: Multi-Kernel Kronecker Adaptation with Re-Scaling Transmission for Visual Classification
cs.CV cs.AI
Fine-tuning pre-trained vision models for specific tasks is a common practice in computer vision. However, this process becomes more expensive as models grow larger. Recently, parameter-efficient fine-tuning (PEFT) methods have emerged as a popular solution to improve training efficiency and reduce storage needs by tuning additional low-rank modules within pre-trained backbones. Despite their advantages, they struggle with limited representation capabilities and misalignment with pre-trained intermediate features. To address these issues, we introduce an innovative Multi-Kernel Kronecker Adaptation with Re-Scaling Transmission (KARST) for various recognition tasks. Specifically, its multi-kernel design extends Kronecker projections horizontally and separates adaptation matrices into multiple complementary spaces, reducing parameter dependency and creating more compact subspaces. Besides, it incorporates extra learnable re-scaling factors to better align with pre-trained feature distributions, allowing for more flexible and balanced feature aggregation. Extensive experiments validate that our KARST outperforms other PEFT counterparts with a negligible inference cost due to its re-parameterization characteristics. Code is publicly available at: https://github.com/Lucenova/KARST.
2502.06781
Exploring the Limit of Outcome Reward for Learning Mathematical Reasoning
cs.CL cs.LG
Reasoning abilities, especially those for solving complex math problems, are crucial components of general intelligence. Recent advances by proprietary companies, such as o-series models of OpenAI, have made remarkable progress on reasoning tasks. However, the complete technical details remain unrevealed, and the techniques that are believed certainly to be adopted are only reinforcement learning (RL) and the long chain of thoughts. This paper proposes a new RL framework, termed OREAL, to pursue the performance limit that can be achieved through \textbf{O}utcome \textbf{RE}w\textbf{A}rd-based reinforcement \textbf{L}earning for mathematical reasoning tasks, where only binary outcome rewards are easily accessible. We theoretically prove that behavior cloning on positive trajectories from best-of-N (BoN) sampling is sufficient to learn the KL-regularized optimal policy in binary feedback environments. This formulation further implies that the rewards of negative samples should be reshaped to ensure the gradient consistency between positive and negative samples. To alleviate the long-existing difficulties brought by sparse rewards in RL, which are even exacerbated by the partial correctness of the long chain of thought for reasoning tasks, we further apply a token-level reward model to sample important tokens in reasoning trajectories for learning. With OREAL, for the first time, a 7B model can obtain 94.0 pass@1 accuracy on MATH-500 through RL, being on par with 32B models. OREAL-32B also surpasses previous 32B models trained by distillation with 95.0 pass@1 accuracy on MATH-500. Our investigation also indicates the importance of initial policy models and training queries for RL. Code, models, and data will be released to benefit future research\footnote{https://github.com/InternLM/OREAL}.
2502.06782
Lumina-Video: Efficient and Flexible Video Generation with Multi-scale Next-DiT
cs.CV
Recent advancements have established Diffusion Transformers (DiTs) as a dominant framework in generative modeling. Building on this success, Lumina-Next achieves exceptional performance in the generation of photorealistic images with Next-DiT. However, its potential for video generation remains largely untapped, with significant challenges in modeling the spatiotemporal complexity inherent to video data. To address this, we introduce Lumina-Video, a framework that leverages the strengths of Next-DiT while introducing tailored solutions for video synthesis. Lumina-Video incorporates a Multi-scale Next-DiT architecture, which jointly learns multiple patchifications to enhance both efficiency and flexibility. By incorporating the motion score as an explicit condition, Lumina-Video also enables direct control of generated videos' dynamic degree. Combined with a progressive training scheme with increasingly higher resolution and FPS, and a multi-source training scheme with mixed natural and synthetic data, Lumina-Video achieves remarkable aesthetic quality and motion smoothness at high training and inference efficiency. We additionally propose Lumina-V2A, a video-to-audio model based on Next-DiT, to create synchronized sounds for generated videos. Codes are released at https://www.github.com/Alpha-VLLM/Lumina-Video.
2502.06784
RelGNN: Composite Message Passing for Relational Deep Learning
cs.LG cs.AI cs.DB
Predictive tasks on relational databases are critical in real-world applications spanning e-commerce, healthcare, and social media. To address these tasks effectively, Relational Deep Learning (RDL) encodes relational data as graphs, enabling Graph Neural Networks (GNNs) to exploit relational structures for improved predictions. However, existing heterogeneous GNNs often overlook the intrinsic structural properties of relational databases, leading to modeling inefficiencies. Here we introduce RelGNN, a novel GNN framework specifically designed to capture the unique characteristics of relational databases. At the core of our approach is the introduction of atomic routes, which are sequences of nodes forming high-order tripartite structures. Building upon these atomic routes, RelGNN designs new composite message passing mechanisms between heterogeneous nodes, allowing direct single-hop interactions between them. This approach avoids redundant aggregations and mitigates information entanglement, ultimately leading to more efficient and accurate predictive modeling. RelGNN is evaluated on 30 diverse real-world tasks from RelBench (Fey et al., 2024), and consistently achieves state-of-the-art accuracy with up to 25% improvement.
2502.06785
DeepCrossAttention: Supercharging Transformer Residual Connections
cs.LG
Transformer networks have achieved remarkable success across diverse domains, leveraging a variety of architectural innovations, including residual connections. However, traditional residual connections, which simply sum the outputs of previous layers, can dilute crucial information. This work introduces DeepCrossAttention (DCA), an approach that enhances residual learning in transformers. DCA employs learnable, input-dependent weights to dynamically combine layer outputs, enabling the model to selectively focus on the most relevant information in any of the previous layers. Furthermore, DCA incorporates depth-wise cross-attention, allowing for richer interactions between layers at different depths. Our language modeling experiments show that DCA achieves improved perplexity for a given training time. Moreover, DCA obtains the same model quality up to 3x faster while adding a negligible number of parameters. Theoretical analysis confirms that DCA provides an improved trade-off between accuracy and model size when the ratio of collective layer ranks to the ambient dimension falls below a critical threshold.
2502.06786
Matryoshka Quantization
cs.LG cs.AI
Quantizing model weights is critical for reducing the communication and inference costs of large models. However, quantizing models -- especially to low precisions like int4 or int2 -- requires a trade-off in model quality; int2, in particular, is known to severely degrade model quality. Consequently, practitioners are often forced to maintain multiple models with different quantization levels or serve a single model that best satisfies the quality-latency trade-off. On the other hand, integer data types, such as int8, inherently possess a nested (Matryoshka) structure where smaller bit-width integers, like int4 or int2, are nested within the most significant bits. This paper proposes Matryoshka Quantization (MatQuant), a novel multi-scale quantization technique that addresses the challenge of needing multiple quantized models. It allows training and maintaining just one model, which can then be served at different precision levels. Furthermore, due to the co-training and co-distillation regularization provided by MatQuant, the int2 precision models extracted by MatQuant can be up to $10\%$ more accurate than standard int2 quantization (using techniques like QAT or OmniQuant). This represents significant progress in model quantization, demonstrated by the fact that, with the same recipe, an int2 FFN-quantized Gemma-2 9B model is more accurate than an int8 FFN-quantized Gemma-2 2B model.
2502.06787
Visual Agentic AI for Spatial Reasoning with a Dynamic API
cs.CV
Visual reasoning -- the ability to interpret the visual world -- is crucial for embodied agents that operate within three-dimensional scenes. Progress in AI has led to vision and language models capable of answering questions from images. However, their performance declines when tasked with 3D spatial reasoning. To tackle the complexity of such reasoning problems, we introduce an agentic program synthesis approach where LLM agents collaboratively generate a Pythonic API with new functions to solve common subproblems. Our method overcomes limitations of prior approaches that rely on a static, human-defined API, allowing it to handle a wider range of queries. To assess AI capabilities for 3D understanding, we introduce a new benchmark of queries involving multiple steps of grounding and inference. We show that our method outperforms prior zero-shot models for visual reasoning in 3D and empirically validate the effectiveness of our agentic framework for 3D spatial reasoning tasks. Project website: https://glab-caltech.github.io/vadar/
2502.06788
EVEv2: Improved Baselines for Encoder-Free Vision-Language Models
cs.CV cs.AI
Existing encoder-free vision-language models (VLMs) are rapidly narrowing the performance gap with their encoder-based counterparts, highlighting the promising potential for unified multimodal systems with structural simplicity and efficient deployment. We systematically clarify the performance gap between VLMs using pre-trained vision encoders, discrete tokenizers, and minimalist visual layers from scratch, deeply excavating the under-examined characteristics of encoder-free VLMs. We develop efficient strategies for encoder-free VLMs that rival mainstream encoder-based ones. After an in-depth investigation, we launch EVEv2.0, a new and improved family of encoder-free VLMs. We show that: (i) Properly decomposing and hierarchically associating vision and language within a unified model reduces interference between modalities. (ii) A well-designed training strategy enables effective optimization for encoder-free VLMs. Through extensive evaluation, our EVEv2.0 represents a thorough study for developing a decoder-only architecture across modalities, demonstrating superior data efficiency and strong vision-reasoning capability. Code is publicly available at: https://github.com/baaivision/EVE.
2502.06789
Information-theoretic Bayesian Optimization: Survey and Tutorial
cs.LG cs.AI cs.IT math.IT stat.ML
Several scenarios require the optimization of non-convex black-box functions, that are noisy expensive to evaluate functions with unknown analytical expression, whose gradients are hence not accessible. For example, the hyper-parameter tuning problem of machine learning models. Bayesian optimization is a class of methods with state-of-the-art performance delivering a solution to this problem in real scenarios. It uses an iterative process that employs a probabilistic surrogate model, typically a Gaussian process, of the objective function to be optimized computing a posterior predictive distribution of the black-box function. Based on the information given by this posterior predictive distribution, Bayesian optimization includes the computation of an acquisition function that represents, for every input space point, the utility of evaluating that point in the next iteraiton if the objective of the process is to retrieve a global extremum. This paper is a survey of the information theoretical acquisition functions, whose performance typically outperforms the rest of acquisition functions. The main concepts of the field of information theory are also described in detail to make the reader aware of why information theory acquisition functions deliver great results in Bayesian optimization and how can we approximate them when they are intractable. We also cover how information theory acquisition functions can be adapted to complex optimization scenarios such as the multi-objective, constrained, non-myopic, multi-fidelity, parallel and asynchronous settings and provide further lines of research.
2502.06798
Prompt-Aware Scheduling for Efficient Text-to-Image Inferencing System
cs.LG cs.DC cs.GR
Traditional ML models utilize controlled approximations during high loads, employing faster, but less accurate models in a process called accuracy scaling. However, this method is less effective for generative text-to-image models due to their sensitivity to input prompts and performance degradation caused by large model loading overheads. This work introduces a novel text-to-image inference system that optimally matches prompts across multiple instances of the same model operating at various approximation levels to deliver high-quality images under high loads and fixed budgets.
2502.06800
Analyzing Geospatial and Socioeconomic Disparities in Breast Cancer Screening Among Populations in the United States: Machine Learning Approach
cs.LG stat.AP
Breast cancer screening plays a pivotal role in early detection and subsequent effective management of the disease, impacting patient outcomes and survival rates. This study aims to assess breast cancer screening rates nationwide in the United States and investigate the impact of social determinants of health on these screening rates. Data on mammography screening at the census tract level for 2018 and 2020 were collected from the Behavioral Risk Factor Surveillance System. We developed a large dataset of social determinants of health, comprising 13 variables for 72337 census tracts. Spatial analysis employing Getis-Ord Gi statistics was used to identify clusters of high and low breast cancer screening rates. To evaluate the influence of these social determinants, we implemented a random forest model, with the aim of comparing its performance to linear regression and support vector machine models. The models were evaluated using R2 and root mean squared error metrics. Shapley Additive Explanations values were subsequently used to assess the significance of variables and direction of their influence. Geospatial analysis revealed elevated screening rates in the eastern and northern United States, while central and midwestern regions exhibited lower rates. The random forest model demonstrated superior performance, with an R2=64.53 and root mean squared error of 2.06 compared to linear regression and support vector machine models. Shapley Additive Explanations values indicated that the percentage of the Black population, the number of mammography facilities within a 10-mile radius, and the percentage of the population with at least a bachelor's degree were the most influential variables, all positively associated with mammography screening rates.
2502.06802
Solving the Content Gap in Roblox Game Recommendations: LLM-Based Profile Generation and Reranking
cs.IR cs.AI cs.CL cs.LG
With the vast and dynamic user-generated content on Roblox, creating effective game recommendations requires a deep understanding of game content. Traditional recommendation models struggle with the inconsistent and sparse nature of game text features such as titles and descriptions. Recent advancements in large language models (LLMs) offer opportunities to enhance recommendation systems by analyzing in-game text data. This paper addresses two challenges: generating high-quality, structured text features for games without extensive human annotation, and validating these features to ensure they improve recommendation relevance. We propose an approach that extracts in-game text and uses LLMs to infer attributes such as genre and gameplay objectives from raw player interactions. Additionally, we introduce an LLM-based re-ranking mechanism to assess the effectiveness of the generated text features, enhancing personalization and user satisfaction. Beyond recommendations, our approach supports applications such as user engagement-based integrity detection, already deployed in production. This scalable framework demonstrates the potential of in-game text understanding to improve recommendation quality on Roblox and adapt recommendations to its unique, user-generated ecosystem.
2502.06803
Emotion Recognition and Generation: A Comprehensive Review of Face, Speech, and Text Modalities
cs.LG cs.AI cs.CV
Emotion recognition and generation have emerged as crucial topics in Artificial Intelligence research, playing a significant role in enhancing human-computer interaction within healthcare, customer service, and other fields. Although several reviews have been conducted on emotion recognition and generation as separate entities, many of these works are either fragmented or limited to specific methodologies, lacking a comprehensive overview of recent developments and trends across different modalities. In this survey, we provide a holistic review aimed at researchers beginning their exploration in emotion recognition and generation. We introduce the fundamental principles underlying emotion recognition and generation across facial, vocal, and textual modalities. This work categorises recent state-of-the-art research into distinct technical approaches and explains the theoretical foundations and motivations behind these methodologies, offering a clearer understanding of their application. Moreover, we discuss evaluation metrics, comparative analyses, and current limitations, shedding light on the challenges faced by researchers in the field. Finally, we propose future research directions to address these challenges and encourage further exploration into developing robust, effective, and ethically responsible emotion recognition and generation systems.
2502.06805
Efficient Diffusion Models: A Survey
cs.LG cs.GR
Diffusion models have emerged as powerful generative models capable of producing high-quality contents such as images, videos, and audio, demonstrating their potential to revolutionize digital content creation. However, these capabilities come at the cost of their significant computational resources and lengthy generation time, underscoring the critical need to develop efficient techniques for practical deployment. In this survey, we provide a systematic and comprehensive review of research on efficient diffusion models. We organize the literature in a taxonomy consisting of three main categories, covering distinct yet interconnected efficient diffusion model topics from algorithm-level, system-level, and framework perspective, respectively. We have also created a GitHub repository where we organize the papers featured in this survey at https://github.com/AIoT-MLSys-Lab/Efficient-Diffusion-Model-Survey. We hope our survey can serve as a valuable resource to help researchers and practitioners gain a systematic understanding of efficient diffusion model research and inspire them to contribute to this important and exciting field.
2502.06806
Logits are All We Need to Adapt Closed Models
cs.LG cs.AI cs.CL
Many commercial Large Language Models (LLMs) are often closed-source, limiting developers to prompt tuning for aligning content generation with specific applications. While these models currently do not provide access to token logits, we argue that if such access were available, it would enable more powerful adaptation techniques beyond prompt engineering. In this paper, we propose a token-level probability reweighting framework that, given access to logits and a small amount of task-specific data, can effectively steer black-box LLMs toward application-specific content generation. Our approach views next-token prediction through the lens of supervised classification. We show that aligning black-box LLMs with task-specific data can be formulated as a label noise correction problem, leading to \emph{Plugin} model -- an autoregressive probability reweighting model that operates solely on logits. We provide theoretical justification for why reweighting logits alone is sufficient for task adaptation. Extensive experiments with multiple datasets, LLMs, and reweighting models demonstrate the effectiveness of our method, advocating for broader access to token logits in closed-source models.
2502.06807
Competitive Programming with Large Reasoning Models
cs.LG cs.AI cs.CL
We show that reinforcement learning applied to large language models (LLMs) significantly boosts performance on complex coding and reasoning tasks. Additionally, we compare two general-purpose reasoning models - OpenAI o1 and an early checkpoint of o3 - with a domain-specific system, o1-ioi, which uses hand-engineered inference strategies designed for competing in the 2024 International Olympiad in Informatics (IOI). We competed live at IOI 2024 with o1-ioi and, using hand-crafted test-time strategies, placed in the 49th percentile. Under relaxed competition constraints, o1-ioi achieved a gold medal. However, when evaluating later models such as o3, we find that o3 achieves gold without hand-crafted domain-specific strategies or relaxed constraints. Our findings show that although specialized pipelines such as o1-ioi yield solid improvements, the scaled-up, general-purpose o3 model surpasses those results without relying on hand-crafted inference heuristics. Notably, o3 achieves a gold medal at the 2024 IOI and obtains a Codeforces rating on par with elite human competitors. Overall, these results indicate that scaling general-purpose reinforcement learning, rather than relying on domain-specific techniques, offers a robust path toward state-of-the-art AI in reasoning domains, such as competitive programming.
2502.06808
On the Benefits of Attribute-Driven Graph Domain Adaptation
cs.LG cs.AI
Graph Domain Adaptation (GDA) addresses a pressing challenge in cross-network learning, particularly pertinent due to the absence of labeled data in real-world graph datasets. Recent studies attempted to learn domain invariant representations by eliminating structural shifts between graphs. In this work, we show that existing methodologies have overlooked the significance of the graph node attribute, a pivotal factor for graph domain alignment. Specifically, we first reveal the impact of node attributes for GDA by theoretically proving that in addition to the graph structural divergence between the domains, the node attribute discrepancy also plays a critical role in GDA. Moreover, we also empirically show that the attribute shift is more substantial than the topology shift, which further underscores the importance of node attribute alignment in GDA. Inspired by this finding, a novel cross-channel module is developed to fuse and align both views between the source and target graphs for GDA. Experimental results on a variety of benchmarks verify the effectiveness of our method.
2502.06809
Neurons Speak in Ranges: Breaking Free from Discrete Neuronal Attribution
cs.LG cs.AI cs.CL
Interpreting and controlling the internal mechanisms of large language models (LLMs) is crucial for improving their trustworthiness and utility. Recent efforts have primarily focused on identifying and manipulating neurons by establishing discrete mappings between neurons and semantic concepts. However, such mappings struggle to handle the inherent polysemanticity in LLMs, where individual neurons encode multiple, distinct concepts. This makes precise control challenging and complicates downstream interventions. Through an in-depth analysis of both encoder and decoder-based LLMs across multiple text classification datasets, we uncover that while individual neurons encode multiple concepts, their activation magnitudes vary across concepts in distinct, Gaussian-like patterns. Building on this insight, we introduce NeuronLens, a novel range-based interpretation and manipulation framework that provides a finer view of neuron activation distributions to localize concept attribution within a neuron. Extensive empirical evaluations demonstrate that NeuronLens significantly reduces unintended interference, while maintaining precise control for manipulation of targeted concepts, outperforming existing methods.
2502.06810
Emergence of Self-Awareness in Artificial Systems: A Minimalist Three-Layer Approach to Artificial Consciousness
q-bio.NC cs.AI
This paper proposes a minimalist three-layer model for artificial consciousness, focusing on the emergence of self-awareness. The model comprises a Cognitive Integration Layer, a Pattern Prediction Layer, and an Instinctive Response Layer, interacting with Access-Oriented and Pattern-Integrated Memory systems. Unlike brain-replication approaches, we aim to achieve minimal self-awareness through essential elements only. Self-awareness emerges from layer interactions and dynamic self-modeling, without initial explicit self-programming. We detail each component's structure, function, and implementation strategies, addressing technical feasibility. This research offers new perspectives on consciousness emergence in artificial systems, with potential implications for human consciousness understanding and adaptable AI development. We conclude by discussing ethical considerations and future research directions.
2502.06811
Aligning Human and Machine Attention for Enhanced Supervised Learning
cs.LG cs.AI cs.CL
Attention, or prioritization of certain information items over others, is a critical element of any learning process, for both humans and machines. Given that humans continue to outperform machines in certain learning tasks, it seems plausible that machine performance could be enriched by aligning machine attention with human attention mechanisms -- yet research on this topic is sparse and has achieved only limited success. This paper proposes a new approach to address this gap, called Human-Machine Attention Learning (HuMAL). This approach involves reliance on data annotated by humans to reflect their self-perceived attention during specific tasks. We evaluate several alternative strategies for integrating such human attention data into machine learning (ML) algorithms, using a sentiment analysis task (review data from Yelp) and a personality-type classification task (data from myPersonality). The best-performing HuMAL strategy significantly enhances the task performance of fine-tuned transformer models (BERT, as well as GPT-2 and XLNET), and the benefit is particularly pronounced under challenging conditions of imbalanced or sparse labeled data. This research contributes to a deeper understanding of strategies for integrating human attention into ML models and highlights the potential of leveraging human cognition to augment ML in real-world applications.
2502.06812
Harness Local Rewards for Global Benefits: Effective Text-to-Video Generation Alignment with Patch-level Reward Models
cs.LG cs.GR
The emergence of diffusion models (DMs) has significantly improved the quality of text-to-video generation models (VGMs). However, current VGM optimization primarily emphasizes the global quality of videos, overlooking localized errors, which leads to suboptimal generation capabilities. To address this issue, we propose a post-training strategy for VGMs, HALO, which explicitly incorporates local feedback from a patch reward model, providing detailed and comprehensive training signals with the video reward model for advanced VGM optimization. To develop an effective patch reward model, we distill GPT-4o to continuously train our video reward model, which enhances training efficiency and ensures consistency between video and patch reward distributions. Furthermore, to harmoniously integrate patch rewards into VGM optimization, we introduce a granular DPO (Gran-DPO) algorithm for DMs, allowing collaborative use of both patch and video rewards during the optimization process. Experimental results indicate that our patch reward model aligns well with human annotations and HALO substantially outperforms the baselines across two evaluation methods. Further experiments quantitatively prove the existence of patch defects, and our proposed method could effectively alleviate this issue.
2502.06813
Policy Guided Tree Search for Enhanced LLM Reasoning
cs.LG cs.AI
Despite their remarkable capabilities, large language models often struggle with tasks requiring complex reasoning and planning. While existing approaches like Chain-of-Thought prompting and tree search techniques show promise, they are limited by their reliance on predefined heuristics and computationally expensive exploration strategies. We propose Policy-Guided Tree Search (PGTS), a framework that combines reinforcement learning with structured tree exploration to efficiently navigate reasoning paths. Our key innovation is a learned policy that dynamically decides between expanding, branching, backtracking, or terminating exploration, eliminating the need for manual heuristics or exhaustive search. Experiments across mathematical reasoning, logical deduction, and planning benchmarks demonstrate that PGTS achieves superior reasoning performance while significantly reducing computational costs compared to existing methods. These results establish PGTS as a scalable and effective solution for tackling complex reasoning tasks with LLMs.
2502.06814
Diffusion Instruction Tuning
cs.LG cs.AI cs.GR
We introduce Lavender, a simple supervised fine-tuning (SFT) method that boosts the performance of advanced vision-language models (VLMs) by leveraging state-of-the-art image generation models such as Stable Diffusion. Specifically, Lavender aligns the text-vision attention in the VLM transformer with the equivalent used by Stable Diffusion during SFT, instead of adapting separate encoders. This alignment enriches the model's visual understanding and significantly boosts performance across in- and out-of-distribution tasks. Lavender requires just 0.13 million training examples, 2.5% of typical large-scale SFT datasets, and fine-tunes on standard hardware (8 GPUs) in a single day. It consistently improves state-of-the-art open-source multimodal LLMs (e.g., Llama-3.2-11B, MiniCPM-Llama3-v2.5), achieving up to 30% gains and a 68% boost on challenging out-of-distribution medical QA tasks. By efficiently transferring the visual expertise of image generators with minimal supervision, Lavender offers a scalable solution for more accurate vision-language systems. All code, training data, and models will be shared at https://astrazeneca.github.io/vlm/.
2502.06815
Honegumi: An Interface for Accelerating the Adoption of Bayesian Optimization in the Experimental Sciences
cs.LG cond-mat.mtrl-sci
Bayesian optimization (BO) has emerged as a powerful tool for guiding experimental design and decision-making in various scientific fields, including materials science, chemistry, and biology. However, despite its growing popularity, the complexity of existing BO libraries and the steep learning curve associated with them can deter researchers who are not well-versed in machine learning or programming. To address this barrier, we introduce Honegumi, a user-friendly, interactive tool designed to simplify the process of creating advanced Bayesian optimization scripts. Honegumi offers a dynamic selection grid that allows users to configure key parameters of their optimization tasks, generating ready-to-use, unit-tested Python scripts tailored to their specific needs. Accompanying the interface is a comprehensive suite of tutorials that provide both conceptual and practical guidance, bridging the gap between theoretical understanding and practical implementation. Built on top of the Ax platform, Honegumi leverages the power of existing state-of-the-art libraries while restructuring the user experience to make advanced BO techniques more accessible to experimental researchers. By lowering the barrier to entry and providing educational resources, Honegumi aims to accelerate the adoption of advanced Bayesian optimization methods across various domains.
2502.06816
DeepCell: Multiview Representation Learning for Post-Mapping Netlists
cs.LG cs.AI
Representation learning for post-mapping (PM) netlists is a critical challenge in Electronic Design Automation (EDA), driven by the diverse and complex nature of modern circuit designs. Existing approaches focus on intermediate representations like And-Inverter Graphs (AIGs), limiting their applicability to post-synthesis stages. We introduce DeepCell, a multiview representation learning framework that integrates structural and functional insights from both PM netlists and AIGs to learn rich, generalizable embeddings. At its core, DeepCell employs the novel Mask Circuit Modeling (MCM) mechanism, which refines PM netlist representations in a self-supervised manner using pretrained AIG encoders. DeepCell sets a new benchmark in PM netlist representation, outperforming existing methods in predictive accuracy and reconstruction fidelity. To validate its efficacy, we apply DeepCell to functional Engineering Change Orders (ECO), achieving significant reductions in patch generation costs and runtime while improving patch quality.
2502.06817
Diffusion-empowered AutoPrompt MedSAM
eess.IV cs.GR cs.LG
MedSAM, a medical foundation model derived from the SAM architecture, has demonstrated notable success across diverse medical domains. However, its clinical application faces two major challenges: the dependency on labor-intensive manual prompt generation, which imposes a significant burden on clinicians, and the absence of semantic labeling in the generated segmentation masks for organs or lesions, limiting its practicality for non-expert users. To address these limitations, we propose AutoMedSAM, an end-to-end framework derived from SAM, designed to enhance usability and segmentation performance. AutoMedSAM retains MedSAM's image encoder and mask decoder structure while introducing a novel diffusion-based class prompt encoder. The diffusion-based encoder employs a dual-decoder structure to collaboratively generate prompt embeddings guided by sparse and dense prompt definitions. These embeddings enhance the model's ability to understand and process clinical imagery autonomously. With this encoder, AutoMedSAM leverages class prompts to embed semantic information into the model's predictions, transforming MedSAM's semi-automated pipeline into a fully automated workflow. Furthermore, AutoMedSAM employs an uncertainty-aware joint optimization strategy during training to effectively inherit MedSAM's pre-trained knowledge while improving generalization by integrating multiple loss functions. Experimental results across diverse datasets demonstrate that AutoMedSAM achieves superior performance while broadening its applicability to both clinical settings and non-expert users. Code is available at https://github.com/HP-ML/AutoPromptMedSAM.git.
2502.06818
Globality Strikes Back: Rethinking the Global Knowledge of CLIP in Training-Free Open-Vocabulary Semantic Segmentation
cs.LG
Recent works modify CLIP to perform open-vocabulary semantic segmentation in a training-free manner (TF-OVSS). In CLIP, patch-wise image representations mainly encode the homogeneous image-level properties and thus are not discriminative enough, hindering its application to the dense prediction task. Previous works make image features more distinct across patches, through making each patch mainly attend to itself or the neighboring patches within a narrow local window. However, with their modifications, the ability of CLIP to aggregate global context information, which is known to be useful for distinguishing confusing categories, is largely weakened. In this paper, we propose a new method named GCLIP, which mines the beneficial global knowledge of CLIP to facilitate the TF-OVSS task. Firstly, we aim to equip the last-block attention with image-level properties while not introducing homogeneous attention patterns across patches. In GCLIP, we merge the attention from the global token emerging blocks with the Query-Query attention to realize this goal. Secondly, we aim to make the Value embeddings of the last-block attention module more distinct and semantically correlated. To realize this, we design a novel channel suppression strategy. As the representation of each patch is finally determined by the attention weights and the Value embeddings, our method can generate more discriminative patch-level image features while absorbing global context information. Extensive experiments on five standard benchmarks demonstrate that our method consistently outperforms previous state-of-the-arts.
2502.06819
Functional 3D Scene Synthesis through Human-Scene Optimization
cs.LG cs.GR
This paper presents a novel generative approach that outputs 3D indoor environments solely from a textual description of the scene. Current methods often treat scene synthesis as a mere layout prediction task, leading to rooms with overlapping objects or overly structured scenes, with limited consideration of the practical usability of the generated environment. Instead, our approach is based on a simple, but effective principle: we condition scene synthesis to generate rooms that are usable by humans. This principle is implemented by synthesizing 3D humans that interact with the objects composing the scene. If this human-centric scene generation is viable, the room layout is functional and it leads to a more coherent 3D structure. To this end, we propose a novel method for functional 3D scene synthesis, which consists of reasoning, 3D assembling and optimization. We regard text guided 3D synthesis as a reasoning process by generating a scene graph via a graph diffusion network. Considering object functional co-occurrence, a new strategy is designed to better accommodate human-object interaction and avoidance, achieving human-aware 3D scene optimization. We conduct both qualitative and quantitative experiments to validate the effectiveness of our method in generating coherent 3D scene synthesis results.
2502.06820
LoCA: Location-Aware Cosine Adaptation for Parameter-Efficient Fine-Tuning
cs.LG cs.AI
Low-rank adaptation (LoRA) has become a prevalent method for adapting pre-trained large language models to downstream tasks. However, the simple low-rank decomposition form may constrain the hypothesis space. To address this limitation, we introduce Location-aware Cosine Adaptation (LoCA), a novel frequency-domain parameter-efficient fine-tuning method based on inverse Discrete Cosine Transform (iDCT) with selective locations of learnable components. We begin with a comprehensive theoretical comparison between frequency-domain and low-rank decompositions for fine-tuning pre-trained large models. Our analysis reveals that frequency-domain approximation with carefully selected frequency components can surpass the expressivity of traditional low-rank-based methods. Furthermore, we demonstrate that iDCT offers a more efficient implementation compared to inverse Discrete Fourier Transform (iDFT), allowing for better selection and tuning of frequency components while maintaining equivalent expressivity to the optimal iDFT-based adaptation. By employing finite-difference approximation to estimate gradients for discrete locations of learnable coefficients on the DCT spectrum, LoCA dynamically selects the most informative frequency components during training. Experiments on diverse language and vision fine-tuning tasks demonstrate that LoCA offers enhanced parameter efficiency while maintains computational feasibility comparable to low-rank-based methods.
2502.06822
DiffListener: Discrete Diffusion Model for Listener Generation
cs.LG cs.CL cs.GR
The listener head generation (LHG) task aims to generate natural nonverbal listener responses based on the speaker's multimodal cues. While prior work either rely on limited modalities (e.g. audio and facial information) or employ autoregressive approaches which have limitations such as accumulating prediction errors. To address these limitations, we propose DiffListener, a discrete diffusion based approach for non-autoregressive listener head generation. Our model takes the speaker's facial information, audio, and text as inputs, additionally incorporating facial differential information to represent the temporal dynamics of expressions and movements. With this explicit modeling of facial dynamics, DiffListener can generate coherent reaction sequences in a non-autoregressive manner. Through comprehensive experiments, DiffListener demonstrates state-of-the-art performance in both quantitative and qualitative evaluations. The user study shows that DiffListener generates natural context-aware listener reactions that are well synchronized with the speaker. The code and demo videos are available in https://siyeoljung.github.io/DiffListener
2502.06823
CTR-Driven Advertising Image Generation with Multimodal Large Language Models
cs.LG cs.CV cs.GR cs.IR
In web data, advertising images are crucial for capturing user attention and improving advertising effectiveness. Most existing methods generate background for products primarily focus on the aesthetic quality, which may fail to achieve satisfactory online performance. To address this limitation, we explore the use of Multimodal Large Language Models (MLLMs) for generating advertising images by optimizing for Click-Through Rate (CTR) as the primary objective. Firstly, we build targeted pre-training tasks, and leverage a large-scale e-commerce multimodal dataset to equip MLLMs with initial capabilities for advertising image generation tasks. To further improve the CTR of generated images, we propose a novel reward model to fine-tune pre-trained MLLMs through Reinforcement Learning (RL), which can jointly utilize multimodal features and accurately reflect user click preferences. Meanwhile, a product-centric preference optimization strategy is developed to ensure that the generated background content aligns with the product characteristics after fine-tuning, enhancing the overall relevance and effectiveness of the advertising images. Extensive experiments have demonstrated that our method achieves state-of-the-art performance in both online and offline metrics. Our code and pre-trained models are publicly available at: https://github.com/Chenguoz/CAIG.
2502.06824
Neural Network-based Vehicular Channel Estimation Performance: Effect of Noise in the Training Set
cs.LG cs.AI
Vehicular communication systems face significant challenges due to high mobility and rapidly changing environments, which affect the channel over which the signals travel. To address these challenges, neural network (NN)-based channel estimation methods have been suggested. These methods are primarily trained on high signal-to-noise ratio (SNR) with the assumption that training a NN in less noisy conditions can result in good generalisation. This study examines the effectiveness of training NN-based channel estimators on mixed SNR datasets compared to training solely on high SNR datasets, as seen in several related works. Estimators evaluated in this work include an architecture that uses convolutional layers and self-attention mechanisms; a method that employs temporal convolutional networks and data pilot-aided estimation; two methods that combine classical methods with multilayer perceptrons; and the current state-of-the-art model that combines Long-Short-Term Memory networks with data pilot-aided and temporal averaging methods as post processing. Our results indicate that using only high SNR data for training is not always optimal, and the SNR range in the training dataset should be treated as a hyperparameter that can be adjusted for better performance. This is illustrated by the better performance of some models in low SNR conditions when trained on the mixed SNR dataset, as opposed to when trained exclusively on high SNR data.
2502.06825
RLOMM: An Efficient and Robust Online Map Matching Framework with Reinforcement Learning
cs.LG cs.DB
Online map matching is a fundamental problem in location-based services, aiming to incrementally match trajectory data step-by-step onto a road network. However, existing methods fail to meet the needs for efficiency, robustness, and accuracy required by large-scale online applications, making this task still a challenging problem. This paper introduces a novel framework that achieves high accuracy and efficient matching while ensuring robustness in handling diverse scenarios. To improve efficiency, we begin by modeling the online map matching problem as an Online Markov Decision Process (OMDP) based on its inherent characteristics. This approach helps efficiently merge historical and real-time data, reducing unnecessary calculations. Next, to enhance the model's robustness, we design a reinforcement learning method, enabling robust handling of real-time data from dynamically changing environments. In particular, we propose a novel model learning process and a comprehensive reward function, allowing the model to make reasonable current matches from a future-oriented perspective, and to continuously update and optimize during the decision-making process based on feedback. Lastly, to address the heterogeneity between trajectories and roads, we design distinct graph structures, facilitating efficient representation learning through graph and recurrent neural networks. To further align trajectory and road data, we introduce contrastive learning to decrease their distance in the latent space, thereby promoting effective integration of the two. Extensive evaluations on three real-world datasets confirm that our method significantly outperforms existing state-of-the-art solutions in terms of accuracy, efficiency and robustness.
2502.06826
Transferring Graph Neural Networks for Soft Sensor Modeling using Process Topologies
cs.LG cs.AI
Data-driven soft sensors help in process operations by providing real-time estimates of otherwise hard- to-measure process quantities, e.g., viscosities or product concentrations. Currently, soft sensors need to be developed individually per plant. Using transfer learning, machine learning-based soft sensors could be reused and fine-tuned across plants and applications. However, transferring data-driven soft sensor models is in practice often not possible, because the fixed input structure of standard soft sensor models prohibits transfer if, e.g., the sensor information is not identical in all plants. We propose a topology-aware graph neural network approach for transfer learning of soft sensor models across multiple plants. In our method, plants are modeled as graphs: Unit operations are nodes, streams are edges, and sensors are embedded as attributes. Our approach brings two advantages for transfer learning: First, we not only include sensor data but also crucial information on the plant topology. Second, the graph neural network algorithm is flexible with respect to its sensor inputs. This allows us to model data from different plants with different sensor networks. We test the transfer learning capabilities of our modeling approach on ammonia synthesis loops with different process topologies. We build a soft sensor predicting the ammonia concentration in the product. After training on data from one process, we successfully transfer our soft sensor model to a previously unseen process with a different topology. Our approach promises to extend the data-driven soft sensors to cases to leverage data from multiple plants.
2502.06827
Learning to Synthesize Compatible Fashion Items Using Semantic Alignment and Collocation Classification: An Outfit Generation Framework
cs.LG cs.AI cs.GR
The field of fashion compatibility learning has attracted great attention from both the academic and industrial communities in recent years. Many studies have been carried out for fashion compatibility prediction, collocated outfit recommendation, artificial intelligence (AI)-enabled compatible fashion design, and related topics. In particular, AI-enabled compatible fashion design can be used to synthesize compatible fashion items or outfits in order to improve the design experience for designers or the efficacy of recommendations for customers. However, previous generative models for collocated fashion synthesis have generally focused on the image-to-image translation between fashion items of upper and lower clothing. In this paper, we propose a novel outfit generation framework, i.e., OutfitGAN, with the aim of synthesizing a set of complementary items to compose an entire outfit, given one extant fashion item and reference masks of target synthesized items. OutfitGAN includes a semantic alignment module, which is responsible for characterizing the mapping correspondence between the existing fashion items and the synthesized ones, to improve the quality of the synthesized images, and a collocation classification module, which is used to improve the compatibility of a synthesized outfit. In order to evaluate the performance of our proposed models, we built a large-scale dataset consisting of 20,000 fashion outfits. Extensive experimental results on this dataset show that our OutfitGAN can synthesize photo-realistic outfits and outperform state-of-the-art methods in terms of similarity, authenticity and compatibility measurements.
2502.06828
Fine-Tuning Strategies for Continual Online EEG Motor Imagery Decoding: Insights from a Large-Scale Longitudinal Study
cs.LG cs.AI
This study investigates continual fine-tuning strategies for deep learning in online longitudinal electroencephalography (EEG) motor imagery (MI) decoding within a causal setting involving a large user group and multiple sessions per participant. We are the first to explore such strategies across a large user group, as longitudinal adaptation is typically studied in the single-subject setting with a single adaptation strategy, which limits the ability to generalize findings. First, we examine the impact of different fine-tuning approaches on decoder performance and stability. Building on this, we integrate online test-time adaptation (OTTA) to adapt the model during deployment, complementing the effects of prior fine-tuning. Our findings demonstrate that fine-tuning that successively builds on prior subject-specific information improves both performance and stability, while OTTA effectively adapts the model to evolving data distributions across consecutive sessions, enabling calibration-free operation. These results offer valuable insights and recommendations for future research in longitudinal online MI decoding and highlight the importance of combining domain adaptation strategies for improving BCI performance in real-world applications. Clinical Relevance: Our investigation enables more stable and efficient long-term motor imagery decoding, which is critical for neurorehabilitation and assistive technologies.
2502.06829
Convolution-Based Converter : A Weak-Prior Approach For Modeling Stochastic Processes Based On Conditional Density Estimation
cs.LG cs.AI
In this paper, a Convolution-Based Converter (CBC) is proposed to develop a methodology for removing the strong or fixed priors in estimating the probability distribution of targets based on observations in the stochastic process. Traditional approaches, e.g., Markov-based and Gaussian process-based methods, typically leverage observations to estimate targets based on strong or fixed priors (such as Markov properties or Gaussian prior). However, the effectiveness of these methods depends on how well their prior assumptions align with the characteristics of the problem. When the assumed priors are not satisfied, these approaches may perform poorly or even become unusable. To overcome the above limitation, we introduce the Convolution-Based converter (CBC), which implicitly estimates the conditional probability distribution of targets without strong or fixed priors, and directly outputs the expected trajectory of the stochastic process that satisfies the constraints from observations. This approach reduces the dependence on priors, enhancing flexibility and adaptability in modeling stochastic processes when addressing different problems. Experimental results demonstrate that our method outperforms existing baselines across multiple metrics.
2502.06830
OrderFusion: Encoding Orderbook for Probabilistic Intraday Price Prediction
q-fin.CP cs.AI cs.LG
Efficient and reliable probabilistic prediction of intraday electricity prices is essential to manage market uncertainties and support robust trading strategies. However, current methods often suffer from parameter inefficiencies, as they fail to fully exploit the potential of modeling interdependencies between bids and offers in the orderbook, requiring a large number of parameters for representation learning. Furthermore, these methods face the quantile crossing issue, where upper quantiles fall below the lower quantiles, resulting in unreliable probabilistic predictions. To address these two challenges, we propose an encoding method called OrderFusion and design a hierarchical multi-quantile head. The OrderFusion encodes the orderbook into a 2.5D representation, which is processed by a tailored jump cross-attention backbone to capture the interdependencies of bids and offers, enabling parameter-efficient learning. The head sets the median quantile as an anchor and predicts multiple quantiles hierarchically, ensuring reliability by enforcing monotonicity between quantiles through non-negative functions. Extensive experiments and ablation studies are conducted on four price indices: 60-min ID3, 60-min ID1, 15-min ID3, and 15-min ID1 using the German orderbook over three years to ensure a fair evaluation. The results confirm that our design choices improve overall performance, offering a parameter-efficient and reliable solution for probabilistic intraday price prediction.
2502.06831
No Location Left Behind: Measuring and Improving the Fairness of Implicit Representations for Earth Data
cs.LG cs.AI
Implicit neural representations (INRs) exhibit growing promise in addressing Earth representation challenges, ranging from emissions monitoring to climate modeling. However, existing methods disproportionately prioritize global average performance, whereas practitioners require fine-grained insights to understand biases and variations in these models. To bridge this gap, we introduce FAIR-Earth: a first-of-its-kind dataset explicitly crafted to examine and challenge inequities in Earth representations. FAIR-Earth comprises various high-resolution Earth signals and uniquely aggregates extensive metadata along stratifications like landmass size and population density to assess the fairness of models. Evaluating state-of-the-art INRs across the various modalities of FAIR-Earth, we uncover striking performance disparities. Certain subgroups, especially those associated with high-frequency signals (e.g., islands, coastlines), are consistently poorly modeled by existing methods. In response, we propose spherical wavelet encodings, building on previous spatial encoding research. Leveraging the multi-resolution capabilities of wavelets, our encodings yield consistent performance over various scales and locations, offering more accurate and robust representations of the biased subgroups. These open-source contributions represent a crucial step towards the equitable assessment and deployment of Earth INRs.
2502.06832
Optimizing Robustness and Accuracy in Mixture of Experts: A Dual-Model Approach
cs.LG cs.AI
Mixture of Experts (MoE) have shown remarkable success in leveraging specialized expert networks for complex machine learning tasks. However, their susceptibility to adversarial attacks presents a critical challenge for deployment in robust applications. This paper addresses the critical question of how to incorporate robustness into MoEs while maintaining high natural accuracy. We begin by analyzing the vulnerability of MoE components, finding that expert networks are notably more susceptible to adversarial attacks than the router. Based on this insight, we propose a targeted robust training technique that integrates a novel loss function to enhance the adversarial robustness of MoE, requiring only the robustification of one additional expert without compromising training or inference efficiency. Building on this, we introduce a dual-model strategy that linearly combines a standard MoE model with our robustified MoE model using a smoothing parameter. This approach allows for flexible control over the robustness-accuracy trade-off. We further provide theoretical foundations by deriving certified robustness bounds for both the single MoE and the dual-model. To push the boundaries of robustness and accuracy, we propose a novel joint training strategy JTDMoE for the dual-model. This joint training enhances both robustness and accuracy beyond what is achievable with separate models. Experimental results on CIFAR-10 and TinyImageNet datasets using ResNet18 and Vision Transformer (ViT) architectures demonstrate the effectiveness of our proposed methods.
2502.06833
Entropy Adaptive Decoding: Dynamic Model Switching for Efficient Inference
cs.LG cs.AI cs.CL
We present Entropy Adaptive Decoding (EAD), a novel approach for efficient language model inference that dynamically switches between different-sized models based on prediction uncertainty. By monitoring rolling entropy in model logit distributions, our method identifies text regions where a smaller model suffices and switches to a larger model only when prediction uncertainty exceeds a threshold. Unlike speculative decoding approaches that maintain perfect output fidelity through verification, EAD accepts controlled output divergence in exchange for computational efficiency. Our experiments on the MATH benchmark demonstrate remarkable efficiency gains across different model families. Using the LLaMA family, we maintain 96.7\% of the 11B model's performance (50.4\% vs 52.1\%) while using it for only 43\% of tokens, decreasing computational cost by 41.5\%. These gains become more pronounced with larger size differentials in the Qwen family, where we achieve 92.9\% of the 14B model's performance (74.3\% vs 80.0\%) while using it for just 25\% of tokens, decreasing computational cost by 67\%. The consistency of these results across model pairs suggests that language model computation can be significantly optimized by selectively deploying model capacity based on local generation complexity. Our findings indicate that current approaches to model inference may be unnecessarily conservative in their pursuit of perfect output fidelity, and that accepting minor performance trade-offs can enable dramatic reductions in computational costs.
2502.06834
A Unified Knowledge-Distillation and Semi-Supervised Learning Framework to Improve Industrial Ads Delivery Systems
cs.LG cs.AI
Industrial ads ranking systems conventionally rely on labeled impression data, which leads to challenges such as overfitting, slower incremental gain from model scaling, and biases due to discrepancies between training and serving data. To overcome these issues, we propose a Unified framework for Knowledge-Distillation and Semi-supervised Learning (UKDSL) for ads ranking, empowering the training of models on a significantly larger and more diverse datasets, thereby reducing overfitting and mitigating training-serving data discrepancies. We provide detailed formal analysis and numerical simulations on the inherent miscalibration and prediction bias of multi-stage ranking systems, and show empirical evidence of the proposed framework's capability to mitigate those. Compared to prior work, UKDSL can enable models to learn from a much larger set of unlabeled data, hence, improving the performance while being computationally efficient. Finally, we report the successful deployment of UKDSL in an industrial setting across various ranking models, serving users at multi-billion scale, across various surfaces, geological locations, clients, and optimize for various events, which to the best of our knowledge is the first of its kind in terms of the scale and efficiency at which it operates.
2502.06835
Reinforcement Learning on AYA Dyads to Enhance Medication Adherence
cs.LG
Medication adherence is critical for the recovery of adolescents and young adults (AYAs) who have undergone hematopoietic cell transplantation (HCT). However, maintaining adherence is challenging for AYAs after hospital discharge, who experience both individual (e.g. physical and emotional symptoms) and interpersonal barriers (e.g., relational difficulties with their care partner, who is often involved in medication management). To optimize the effectiveness of a three-component digital intervention targeting both members of the dyad as well as their relationship, we propose a novel Multi-Agent Reinforcement Learning (MARL) approach to personalize the delivery of interventions. By incorporating the domain knowledge, the MARL framework, where each agent is responsible for the delivery of one intervention component, allows for faster learning compared with a flattened agent. Evaluation using a dyadic simulator environment, based on real clinical data, shows a significant improvement in medication adherence (approximately 3%) compared to purely random intervention delivery. The effectiveness of this approach will be further evaluated in an upcoming trial.
2502.06836
CAST: Cross Attention based multimodal fusion of Structure and Text for materials property prediction
cs.LG cond-mat.mtrl-sci cs.AI
Recent advancements in AI have revolutionized property prediction in materials science and accelerating material discovery. Graph neural networks (GNNs) stand out due to their ability to represent crystal structures as graphs, effectively capturing local interactions and delivering superior predictions. However, these methods often lose critical global information, such as crystal systems and repetitive unit connectivity. To address this, we propose CAST, a cross-attention-based multimodal fusion model that integrates graph and text modalities to preserve essential material information. CAST combines node- and token-level features using cross-attention mechanisms, surpassing previous approaches reliant on material-level embeddings like graph mean-pooling or [CLS] tokens. A masked node prediction pretraining strategy further enhances atomic-level information integration. Our method achieved up to 22.9\% improvement in property prediction across four crystal properties including band gap compared to methods like CrysMMNet and MultiMat. Pretraining was key to aligning node and text embeddings, with attention maps confirming its effectiveness in capturing relationships between nodes and tokens. This study highlights the potential of multimodal learning in materials science, paving the way for more robust predictive models that incorporate both local and global information.
2502.06837
Comparison of CNN-based deep learning architectures for unsteady CFD acceleration on small datasets
cs.LG physics.flu-dyn
CFD acceleration for virtual nuclear reactors or digital twin technology is a primary goal in the nuclear industry. This study compares advanced convolutional neural network (CNN) architectures for accelerating unsteady computational fluid dynamics (CFD) simulations using small datasets based on a challenging natural convection flow dataset. The advanced architectures such as autoencoders, UNet, and ConvLSTM-UNet, were evaluated under identical conditions to determine their predictive accuracy and robustness in autoregressive time-series predictions. ConvLSTM-UNet consistently outperformed other models, particularly in difference value calculation, achieving lower maximum errors and stable residuals. However, error accumulation remains a challenge, limiting reliable predictions to approximately 10 timesteps. This highlights the need for enhanced strategies to improve long-term prediction stability. The novelty of this work lies in its fair comparison of state-of-the-art CNN models within the RePIT framework, demonstrating their potential for accelerating CFD simulations while identifying limitations under small data conditions. Future research will focus on exploring alternative models, such as graph neural networks and implicit neural representations. These efforts aim to develop a robust hybrid approach for long-term unsteady CFD acceleration, contributing to practical applications in virtual nuclear reactor.
2502.06838
TorchResist: Open-Source Differentiable Resist Simulator
cs.LG
Recent decades have witnessed remarkable advancements in artificial intelligence (AI), including large language models (LLMs), image and video generative models, and embodied AI systems. These advancements have led to an explosive increase in the demand for computational power, challenging the limits of Moore's Law. Optical lithography, a critical technology in semiconductor manufacturing, faces significant challenges due to its high costs. To address this, various lithography simulators have been developed. However, many of these simulators are limited by their inadequate photoresist modeling capabilities. This paper presents TorchResist, an open-source, differentiable photoresist simulator.TorchResist employs an analytical approach to model the photoresist process, functioning as a white-box system with at most twenty interpretable parameters. Leveraging modern differentiable programming techniques and parallel computing on GPUs, TorchResist enables seamless co-optimization with other tools across multiple related tasks. Our experimental results demonstrate that TorchResist achieves superior accuracy and efficiency compared to existing solutions. The source code is publicly available.
2502.06839
A Hybrid Model for Weakly-Supervised Speech Dereverberation
eess.AS cs.AI cs.SD eess.SP
This paper introduces a new training strategy to improve speech dereverberation systems using minimal acoustic information and reverberant (wet) speech. Most existing algorithms rely on paired dry/wet data, which is difficult to obtain, or on target metrics that may not adequately capture reverberation characteristics and can lead to poor results on non-target metrics. Our approach uses limited acoustic information, like the reverberation time (RT60), to train a dereverberation system. The system's output is resynthesized using a generated room impulse response and compared with the original reverberant speech, providing a novel reverberation matching loss replacing the standard target metrics. During inference, only the trained dereverberation model is used. Experimental results demonstrate that our method achieves more consistent performance across various objective metrics used in speech dereverberation than the state-of-the-art.
2502.06842
Integrating Generative Artificial Intelligence in ADRD: A Framework for Streamlining Diagnosis and Care in Neurodegenerative Diseases
cs.CY cs.AI
Healthcare systems are struggling to meet the growing demand for neurological care, with challenges particularly acute in Alzheimer's disease and related dementias (ADRD). While artificial intelligence research has often focused on identifying patterns beyond human perception, implementing such predictive capabilities remains challenging as clinicians cannot readily verify insights they cannot themselves detect. We propose that large language models (LLMs) offer more immediately practical applications by enhancing clinicians' capabilities in three critical areas: comprehensive data collection, interpretation of complex clinical information, and timely application of relevant medical knowledge. These challenges stem from limited time for proper diagnosis, growing data complexity, and an overwhelming volume of medical literature that exceeds any clinician's capacity to fully master. We present a framework for responsible AI integration that leverages LLMs' ability to communicate effectively with both patients and providers while maintaining human oversight. This approach prioritizes standardized, high-quality data collection to enable a system that learns from every patient encounter while incorporating the latest clinical evidence, continuously improving care delivery. We begin to address implementation challenges and initiate important discussions around ethical considerations and governance needs. While developed for ADRD, this roadmap provides principles for responsible AI integration across neurology and other medical specialties, with potential to improve diagnostic accuracy, reduce care disparities, and advance clinical knowledge through a learning healthcare system.
2502.06843
Vision-Integrated LLMs for Autonomous Driving Assistance : Human Performance Comparison and Trust Evaluation
cs.CV cs.AI cs.HC
Traditional autonomous driving systems often struggle with reasoning in complex, unexpected scenarios due to limited comprehension of spatial relationships. In response, this study introduces a Large Language Model (LLM)-based Autonomous Driving (AD) assistance system that integrates a vision adapter and an LLM reasoning module to enhance visual understanding and decision-making. The vision adapter, combining YOLOv4 and Vision Transformer (ViT), extracts comprehensive visual features, while GPT-4 enables human-like spatial reasoning and response generation. Experimental evaluations with 45 experienced drivers revealed that the system closely mirrors human performance in describing situations and moderately aligns with human decisions in generating appropriate responses.
2502.06844
Exploring Model Invariance with Discrete Search for Ultra-Low-Bit Quantization
cs.LG cs.AI cs.CL
Large language models have been increasing in size due to their success in a wide range of applications. This calls for a pressing need to reduce memory usage to make them more accessible. Post-training quantization is a popular technique which uses fewer bits (e.g., 4--8 bits) to represent the model without retraining it. However, it remains a challenging task to perform quantization in an ultra-low-bit setup (e.g., 2 bits). In this paper, we propose InvarExplore, a unified framework that systematically explores different model invariance at the same time, allowing us to take advantage of the synergy between each type of invariance. Importantly, InvarExplore features a discrete search algorithm that enables us to explore permutation invariance, which is under-studied as it cannot be optimized with gradient-based methods. Results show that InvarExplore is compatible with existing state-of-the-art methods, achieving an add-on performance improvement over strong competing methods.
2502.06845
DiffNMR3: Advancing NMR Resolution Beyond Instrumental Limits
physics.ins-det cs.AI cs.LG
Nuclear Magnetic Resonance (NMR) spectroscopy is a crucial analytical technique used for molecular structure elucidation, with applications spanning chemistry, biology, materials science, and medicine. However, the frequency resolution of NMR spectra is limited by the "field strength" of the instrument. High-field NMR instruments provide high-resolution spectra but are prohibitively expensive, whereas lower-field instruments offer more accessible, but lower-resolution, results. This paper introduces an AI-driven approach that not only enhances the frequency resolution of NMR spectra through super-resolution techniques but also provides multi-scale functionality. By leveraging a diffusion model, our method can reconstruct high-field spectra from low-field NMR data, offering flexibility in generating spectra at varying magnetic field strengths. These reconstructions are comparable to those obtained from high-field instruments, enabling finer spectral details and improving molecular characterization. To date, our approach is one of the first to overcome the limitations of instrument field strength, achieving NMR super-resolution through AI. This cost-effective solution makes high-resolution analysis accessible to more researchers and industries, without the need for multimillion-dollar equipment.
2502.06846
Prot2Chat: Protein LLM with Early Fusion of Sequence and Structure
cs.LG cs.AI q-bio.BM
Proteins play a pivotal role in living organisms, yet understanding their functions presents significant challenges, including the limited flexibility of classification-based methods, the inability to effectively leverage spatial structural information, and the lack of systematic evaluation metrics for protein Q&A systems. To address these limitations, we propose Prot2Chat, a novel framework that integrates multimodal protein representations with natural language through a unified module, enabling large language model (LLM)-driven answer generation. Our model incorporates a modified ProteinMPNN encoder, which encodes protein sequence and structural information in a unified manner, a protein-text adapter with cross-attention mechanisms, and a LLaMA3 decoder. To optimize training efficiency, we freeze the encoder and employ LoRA techniques for the decoder. We conducted experiments on two datasets, both automated metrics and expert evaluations demonstrate the superior performance of our model. Furthermore, zero-shot prediction results highlight its strong generalization capabilities. This framework offers a promising solution for bridging protein domain knowledge with natural language understanding, paving the way for transformative advancements in protein-related research.
2502.06847
A Deep Learning Framework Integrating CNN and BiLSTM for Financial Systemic Risk Analysis and Prediction
cs.LG cs.CE
This study proposes a deep learning model based on the combination of convolutional neural network (CNN) and bidirectional long short-term memory network (BiLSTM) for discriminant analysis of financial systemic risk. The model first uses CNN to extract local patterns of multidimensional features of financial markets, and then models the bidirectional dependency of time series through BiLSTM, to comprehensively characterize the changing laws of systemic risk in spatial features and temporal dynamics. The experiment is based on real financial data sets. The results show that the model is significantly superior to traditional single models (such as BiLSTM, CNN, Transformer, and TCN) in terms of accuracy, recall, and F1 score. The F1-score reaches 0.88, showing extremely high discriminant ability. This shows that the joint strategy of combining CNN and BiLSTM can not only fully capture the complex patterns of market data but also effectively deal with the long-term dependency problem in time series data. In addition, this study also explores the robustness of the model in dealing with data noise and processing high-dimensional data, providing strong support for intelligent financial risk management. In the future, the research will further optimize the model structure, introduce methods such as reinforcement learning and multimodal data analysis, and improve the efficiency and generalization ability of the model to cope with a more complex financial environment.
2502.06848
Transfer learning in Scalable Graph Neural Network for Improved Physical Simulation
cs.LG cs.AI
In recent years, Graph Neural Network (GNN) based models have shown promising results in simulating physics of complex systems. However, training dedicated graph network based physics simulators can be costly, as most models are confined to fully supervised training, which requires extensive data generated from traditional physics simulators. To date, how transfer learning could improve the model performance and training efficiency has remained unexplored. In this work, we introduce a pre-training and transfer learning paradigm for graph network simulators. We propose the scalable graph U-net (SGUNET). Incorporating an innovative depth-first search (DFS) pooling, the SGUNET is adaptable to different mesh sizes and resolutions for various simulation tasks. To enable the transfer learning between differently configured SGUNETs, we propose a set of mapping functions to align the parameters between the pre-trained model and the target model. An extra normalization term is also added into the loss to constrain the difference between the pre-trained weights and target model weights for better generalization performance. To pre-train our physics simulator we created a dataset which includes 20,000 physical simulations of randomly selected 3D shapes from the open source A Big CAD (ABC) dataset. We show that our proposed transfer learning methods allow the model to perform even better when fine-tuned with small amounts of training data than when it is trained from scratch with full extensive dataset. On the 2D Deformable Plate benchmark dataset, our pre-trained model fine-tuned on 1/16 of the training data achieved an 11.05\% improvement in position RMSE compared to the model trained from scratch.
2502.06849
Model Fusion via Neuron Transplantation
cs.LG cs.AI
Ensemble learning is a widespread technique to improve the prediction performance of neural networks. However, it comes at the price of increased memory and inference time. In this work we propose a novel model fusion technique called \emph{Neuron Transplantation (NT)} in which we fuse an ensemble of models by transplanting important neurons from all ensemble members into the vacant space obtained by pruning insignificant neurons. An initial loss in performance post-transplantation can be quickly recovered via fine-tuning, consistently outperforming individual ensemble members of the same model capacity and architecture. Furthermore, NT enables all the ensemble members to be jointly pruned and jointly trained in a combined model. Comparing it to alignment-based averaging (like Optimal-Transport-fusion), it requires less fine-tuning than the corresponding OT-fused model, the fusion itself is faster and requires less memory, while the resulting model performance is comparable or better. The code is available under the following link: https://github.com/masterbaer/neuron-transplantation.
2502.06851
Survey on Vision-Language-Action Models
cs.CL cs.AI cs.CV
This paper presents an AI-generated review of Vision-Language-Action (VLA) models, summarizing key methodologies, findings, and future directions. The content is produced using large language models (LLMs) and is intended only for demonstration purposes. This work does not represent original research, but highlights how AI can help automate literature reviews. As AI-generated content becomes more prevalent, ensuring accuracy, reliability, and proper synthesis remains a challenge. Future research will focus on developing a structured framework for AI-assisted literature reviews, exploring techniques to enhance citation accuracy, source credibility, and contextual understanding. By examining the potential and limitations of LLM in academic writing, this study aims to contribute to the broader discussion of integrating AI into research workflows. This work serves as a preliminary step toward establishing systematic approaches for leveraging AI in literature review generation, making academic knowledge synthesis more efficient and scalable.
2502.06852
EAP-GP: Mitigating Saturation Effect in Gradient-based Automated Circuit Identification
cs.LG cs.AI
Understanding the internal mechanisms of transformer-based language models remains challenging. Mechanistic interpretability based on circuit discovery aims to reverse engineer neural networks by analyzing their internal processes at the level of computational subgraphs. In this paper, we revisit existing gradient-based circuit identification methods and find that their performance is either affected by the zero-gradient problem or saturation effects, where edge attribution scores become insensitive to input changes, resulting in noisy and unreliable attribution evaluations for circuit components. To address the saturation effect, we propose Edge Attribution Patching with GradPath (EAP-GP), EAP-GP introduces an integration path, starting from the input and adaptively following the direction of the difference between the gradients of corrupted and clean inputs to avoid the saturated region. This approach enhances attribution reliability and improves the faithfulness of circuit identification. We evaluate EAP-GP on 6 datasets using GPT-2 Small, GPT-2 Medium, and GPT-2 XL. Experimental results demonstrate that EAP-GP outperforms existing methods in circuit faithfulness, achieving improvements up to 17.7%. Comparisons with manually annotated ground-truth circuits demonstrate that EAP-GP achieves precision and recall comparable to or better than previous approaches, highlighting its effectiveness in identifying accurate circuits.
2502.06853
Native Fortran Implementation of TensorFlow-Trained Deep and Bayesian Neural Networks
cs.LG cs.AI
Over the past decade, the investigation of machine learning (ML) within the field of nuclear engineering has grown significantly. With many approaches reaching maturity, the next phase of investigation will determine the feasibility and usefulness of ML model implementation in a production setting. Several of the codes used for reactor design and assessment are primarily written in the Fortran language, which is not immediately compatible with TensorFlow-trained ML models. This study presents a framework for implementing deep neural networks (DNNs) and Bayesian neural networks (BNNs) in Fortran, allowing for native execution without TensorFlow's C API, Python runtime, or ONNX conversion. Designed for ease of use and computational efficiency, the framework can be implemented in any Fortran code, supporting iterative solvers and UQ via ensembles or BNNs. Verification was performed using a two-input, one-output test case composed of a noisy sinusoid to compare Fortran-based predictions to those from TensorFlow. The DNN predictions showed negligible differences and achieved a 19.6x speedup, whereas the BNN predictions exhibited minor disagreement, plausibly due to differences in random number generation. An 8.0x speedup was noted for BNN inference. The approach was then further verified on a nuclear-relevant problem predicting critical heat flux (CHF), which demonstrated similar behavior along with significant computational gains. Discussion regarding the framework's successful integration into the CTF thermal-hydraulics code is also included, outlining its practical usefulness. Overall, this framework was shown to be effective at implementing both DNN and BNN model inference within Fortran, allowing for the continued study of ML-based methods in real-world nuclear applications.
2502.06854
Can Large Language Models Understand Intermediate Representations?
cs.LG cs.AI cs.CL
Intermediate Representations (IRs) are essential in compiler design and program analysis, yet their comprehension by Large Language Models (LLMs) remains underexplored. This paper presents a pioneering empirical study to investigate the capabilities of LLMs, including GPT-4, GPT-3, Gemma 2, LLaMA 3.1, and Code Llama, in understanding IRs. We analyze their performance across four tasks: Control Flow Graph (CFG) reconstruction, decompilation, code summarization, and execution reasoning. Our results indicate that while LLMs demonstrate competence in parsing IR syntax and recognizing high-level structures, they struggle with control flow reasoning, execution semantics, and loop handling. Specifically, they often misinterpret branching instructions, omit critical IR operations, and rely on heuristic-based reasoning, leading to errors in CFG reconstruction, IR decompilation, and execution reasoning. The study underscores the necessity for IR-specific enhancements in LLMs, recommending fine-tuning on structured IR datasets and integration of explicit control flow models to augment their comprehension and handling of IR-related tasks.
2502.06855
Self-Supervised Prompt Optimization
cs.CL cs.AI cs.LG
Well-designed prompts are crucial for enhancing Large language models' (LLMs) reasoning capabilities while aligning their outputs with task requirements across diverse domains. However, manually designed prompts require expertise and iterative experimentation. While existing prompt optimization methods aim to automate this process, they rely heavily on external references such as ground truth or by humans, limiting their applicability in real-world scenarios where such data is unavailable or costly to obtain. To address this, we propose Self-Supervised Prompt Optimization (SPO), a cost-efficient framework that discovers effective prompts for both closed and open-ended tasks without requiring external reference. Motivated by the observations that prompt quality manifests directly in LLM outputs and LLMs can effectively assess adherence to task requirements, we derive evaluation and optimization signals purely from output comparisons. Specifically, SPO selects superior prompts through pairwise output comparisons evaluated by an LLM evaluator, followed by an LLM optimizer that aligns outputs with task requirements. Extensive experiments demonstrate that SPO outperforms state-of-the-art prompt optimization methods, achieving comparable or superior results with significantly lower costs (e.g., 1.1% to 5.6% of existing methods) and fewer samples (e.g., three samples). The code is available at https://github.com/geekan/MetaGPT/blob/main/examples/spo
2502.06857
Gemstones: A Model Suite for Multi-Faceted Scaling Laws
cs.LG cs.AI
Scaling laws are typically fit using a family of models with a narrow range of frozen hyper-parameter choices. In this work we study scaling laws using a wide range of architecture and hyper-parameter choices, and highlight their impact on resulting prescriptions. As a primary artifact of our research, we release the Gemstones: the most comprehensive open-source scaling law dataset to date, consisting of over 4000 checkpoints from transformers with up to 2 billion parameters; these models have been trained with different learning rates, cooldown schedules, and architectural shapes. Our checkpoints enable more complex studies of scaling, such as a law that predicts language modeling performance as a function of model width and depth. By examining the various facets of our model suite, we find that the prescriptions of scaling laws can be highly sensitive to the experimental design process and the specific model checkpoints used during fitting. Code: https://github.com/mcleish7/gemstone-scaling-laws
2502.06858
LLM-Supported Natural Language to Bash Translation
cs.CL cs.AI
The Bourne-Again Shell (Bash) command-line interface for Linux systems has complex syntax and requires extensive specialized knowledge. Using the natural language to Bash command (NL2SH) translation capabilities of large language models (LLMs) for command composition circumvents these issues. However, the NL2SH performance of LLMs is difficult to assess due to inaccurate test data and unreliable heuristics for determining the functional equivalence of Bash commands. We present a manually verified test dataset of 600 instruction-command pairs and a training dataset of 40,939 pairs, increasing the size of previous datasets by 441% and 135%, respectively. Further, we present a novel functional equivalence heuristic that combines command execution with LLM evaluation of command outputs. Our heuristic can determine the functional equivalence of two Bash commands with 95% confidence, a 16% increase over previous heuristics. Evaluation of popular LLMs using our test dataset and heuristic demonstrates that parsing, in-context learning, in-weight learning, and constrained decoding can improve NL2SH accuracy by up to 32%. Our findings emphasize the importance of dataset quality, execution-based evaluation and translation method for advancing NL2SH translation. Our code is available at https://github.com/westenfelder/NL2SH