id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
2501.11361
Block Flow: Learning Straight Flow on Data Blocks
cs.LG cs.CV
Flow-matching models provide a powerful framework for various applications, offering efficient sampling and flexible probability path modeling. These models are characterized by flows with low curvature in learned generative trajectories, which results in reduced truncation error at each sampling step. To further reduce curvature, we propose block matching. This novel approach leverages label information to partition the data distribution into blocks and match them with a prior distribution parameterized using the same label information, thereby learning straighter flows. We demonstrate that the variance of the prior distribution can control the curvature upper bound of forward trajectories in flow-matching models. By designing flexible regularization strategies to adjust this variance, we achieve optimal generation performance, effectively balancing the trade-off between maintaining diversity in generated samples and minimizing numerical solver errors. Our results demonstrate competitive performance with models of the same parameter scale.Code is available at \url{https://github.com/wpp13749/block_flow}.
2501.11371
Reed-Solomon Codes Against Insertions and Deletions: Full-Length and Rate-$1/2$ Codes
cs.IT math.IT
The performance of Reed-Solomon codes (RS codes, for short) in the presence of insertion and deletion errors has been studied recently in several papers. In this work, we further study this intriguing mathematical problem, focusing on two regimes. First, we study the question of how well full-length RS codes perform against insertions and deletions. For 2-dimensional RS codes, we fully characterize which codes cannot correct even a single insertion or deletion and show that (almost) all 2-dimensional RS codes correct at least $1$ insertion or deletion error. Moreover, for large enough field size $q$, and for any $k \ge 2$, we show that there exists a full-length $k$-dimensional RS code that corrects $q/10k$ insertions and deletions. Second, we focus on rate $1/2$ RS codes that can correct a single insertion or deletion error. We present a polynomial time algorithm that constructs such codes for $q = O(k^4)$. This result matches the existential bound given in \cite{con2023reed}.
2501.11374
Linear ADRC is equivalent to PID with set-point weighting and measurement filter
eess.SY cs.SY
We show that linear Active Disturbance-Rejection Control (ADRC) tuned using the "bandwidth method" is equivalent to PI(D) control with set-point weighting and a lowpass filter on the measurement signal. We also provide simple expressions that make it possible to implement linear ADRC for first and second-order systems using commonplace two degree-of-freedom PID implementations. The expressions are equivalent to ADRC in the response from measurements, and a slight approximation in the response from references.
2501.11378
Investigation of Whisper ASR Hallucinations Induced by Non-Speech Audio
cs.SD cs.AI eess.AS
Hallucinations of deep neural models are amongst key challenges in automatic speech recognition (ASR). In this paper, we investigate hallucinations of the Whisper ASR model induced by non-speech audio segments present during inference. By inducting hallucinations with various types of sounds, we show that there exists a set of hallucinations that appear frequently. We then study hallucinations caused by the augmentation of speech with such sounds. Finally, we describe the creation of a bag of hallucinations (BoH) that allows to remove the effect of hallucinations through the post-processing of text transcriptions. The results of our experiments show that such post-processing is capable of reducing word error rate (WER) and acts as a good safeguard against problematic hallucinations.
2501.11384
Transductive Conformal Inference for Ranking
cs.LG stat.ME stat.ML
We introduce a method based on Conformal Prediction (CP) to quantify the uncertainty of full ranking algorithms. We focus on a specific scenario where $n + m$ items are to be ranked by some ''black box'' algorithm. It is assumed that the relative (ground truth) ranking of n of them is known. The objective is then to quantify the error made by the algorithm on the ranks of the m new items among the total $(n + m)$. In such a setting, the true ranks of the n original items in the total $(n + m)$ depend on the (unknown) true ranks of the m new ones. Consequently, we have no direct access to a calibration set to apply a classical CP method. To address this challenge, we propose to construct distribution-free bounds of the unknown conformity scores using recent results on the distribution of conformal p-values. Using these scores upper bounds, we provide valid prediction sets for the rank of any item. We also control the false coverage proportion, a crucial quantity when dealing with multiple prediction sets. Finally, we empirically show on both synthetic and real data the efficiency of our CP method.
2501.11388
UniTrans: A Unified Vertical Federated Knowledge Transfer Framework for Enhancing Cross-Hospital Collaboration
cs.LG cs.DC
Cross-hospital collaboration has the potential to address disparities in medical resources across different regions. However, strict privacy regulations prohibit the direct sharing of sensitive patient information between hospitals. Vertical federated learning (VFL) offers a novel privacy-preserving machine learning paradigm that maximizes data utility across multiple hospitals. Traditional VFL methods, however, primarily benefit patients with overlapping data, leaving vulnerable non-overlapping patients without guaranteed improvements in medical prediction services. While some knowledge transfer techniques can enhance the prediction performance for non-overlapping patients, they fall short in addressing scenarios where overlapping and non-overlapping patients belong to different domains, resulting in challenges such as feature heterogeneity and label heterogeneity. To address these issues, we propose a novel unified vertical federated knowledge transfer framework (Unitrans). Our framework consists of three key steps. First, we extract the federated representation of overlapping patients by employing an effective vertical federated representation learning method to model multi-party joint features online. Next, each hospital learns a local knowledge transfer module offline, enabling the transfer of knowledge from the federated representation of overlapping patients to the enriched representation of local non-overlapping patients in a domain-adaptive manner. Finally, hospitals utilize these enriched local representations to enhance performance across various downstream medical prediction tasks. Experiments on real-world medical datasets validate the framework's dual effectiveness in both intra-domain and cross-domain knowledge transfer. The code of \method is available at \url{https://github.com/Chung-ju/Unitrans}.
2501.11391
Revisiting Language Models in Neural News Recommender Systems
cs.IR
Neural news recommender systems (RSs) have integrated language models (LMs) to encode news articles with rich textual information into representations, thereby improving the recommendation process. Most studies suggest that (i) news RSs achieve better performance with larger pre-trained language models (PLMs) than shallow language models (SLMs), and (ii) that large language models (LLMs) outperform PLMs. However, other studies indicate that PLMs sometimes lead to worse performance than SLMs. Thus, it remains unclear whether using larger LMs consistently improves the performance of news RSs. In this paper, we revisit, unify, and extend these comparisons of the effectiveness of LMs in news RSs using the real-world MIND dataset. We find that (i) larger LMs do not necessarily translate to better performance in news RSs, and (ii) they require stricter fine-tuning hyperparameter selection and greater computational resources to achieve optimal recommendation performance than smaller LMs. On the positive side, our experiments show that larger LMs lead to better recommendation performance for cold-start users: they alleviate dependency on extensive user interaction history and make recommendations more reliant on the news content.
2501.11393
Trace Reconstruction of First-Order Reed-Muller Codewords Using Run Statistics
cs.IT math.IT math.PR
In this paper, we derive an expression for the expected number of runs in a trace of a binary sequence $x \in \{0,1\}^n$ obtained by passing $x$ through a deletion channel that independently deletes each bit with probability $q$. We use this expression to show that if $x$ is a codeword of a first-order Reed-Muller code, and the deletion probability $q$ is 1/2, then $x$ can be reconstructed, with high probability, from $\tilde{O}(n)$ many of its traces.
2501.11395
To BEE or not to BEE: Estimating more than Entropy with Biased Entropy Estimators
cs.IT cs.SE math.IT
Entropy estimation plays a significant role in biology, economics, physics, communication engineering and other disciplines. It is increasingly used in software engineering, e.g. in software confidentiality, software testing, predictive analysis, machine learning, and software improvement. However accurate estimation is demonstrably expensive in many contexts, including software. Statisticians have consequently developed biased estimators that aim to accurately estimate entropy on the basis of a sample. In this paper we apply 18 widely employed entropy estimators to Shannon measures useful to the software engineer: entropy, mutual information and conditional mutual information. Moreover, we investigate how the estimators are affected by two main influential factors: sample size and domain size. Our experiments range over a large set of randomly generated joint probability distributions and varying sample sizes, rather than choosing just one or two well known probability distributions as in previous investigations. Our most important result is identifying that the Chao-Shen and Chao-Wang-Jost estimators stand out for consistently converging more quickly to the ground truth, regardless of domain size and regardless of the measure used. They also tend to outperform the others in terms of accuracy as sample sizes increase. This discovery enables a significant reduction in data collection effort without compromising performance.
2501.11403
Verifying Cross-modal Entity Consistency in News using Vision-language Models
cs.CL cs.IR cs.MM
The web has become a crucial source of information, but it is also used to spread disinformation, often conveyed through multiple modalities like images and text. The identification of inconsistent cross-modal information, in particular entities such as persons, locations, and events, is critical to detect disinformation. Previous works either identify out-of-context disinformation by assessing the consistency of images to the whole document, neglecting relations of individual entities, or focus on generic entities that are not relevant to news. So far, only few approaches have addressed the task of validating entity consistency between images and text in news. However, the potential of large vision-language models (LVLMs) has not been explored yet. In this paper, we propose an LVLM-based framework for verifying Cross-modal Entity Consistency~(LVLM4CEC), to assess whether persons, locations and events in news articles are consistent across both modalities. We suggest effective prompting strategies for LVLMs for entity verification that leverage reference images crawled from web. Moreover, we extend three existing datasets for the task of entity verification in news providing manual ground-truth data. Our results show the potential of LVLMs for automating cross-modal entity verification, showing improved accuracy in identifying persons and events when using evidence images. Moreover, our method outperforms a baseline for location and event verification in documents. The datasets and source code are available on GitHub at https://github.com/TIBHannover/LVLM4CEC.
2501.11406
Efficient Reduction of Interconnected Subsystem Models using Abstracted Environments
eess.SY cs.SY
We present two frameworks for structure-preserving model order reduction of interconnected subsystems, improving tractability of the reduction methods while ensuring stability and accuracy bounds of the reduced interconnected model. Instead of reducing each subsystem independently, we take a low-order abstraction of its environment into account to better capture the dynamics relevant to the external input-output behaviour of the interconnected system, thereby increasing accuracy of the reduced interconnected model. This approach significantly reduces the computational costs of reduction by abstracting instead of fully retaining the environment. The two frameworks differ in how they generate these abstracted environments: one abstracts the environment as a whole, whereas the other abstracts each individual subsystem. By relating low-level errors introduced by reduction and abstraction to the resulting high-level error on the interconnected system, we are able to translate high-level accuracy requirements (on the reduced interconnected system) to low-level specifications (on abstraction and reduction errors) using techniques from robust performance analysis. By adhering to these low-level specifications, restricting the introduced low-level errors, both frameworks automatically guarantee the accuracy and stability of the reduced interconnected system. We demonstrate the effectiveness of both frameworks by applying them to a structural dynamics model of a two-stroke wafer stage, achieving improved accuracy and/or greater reduction compared to an existing method from literature.
2501.11407
A Truly Sparse and General Implementation of Gradient-Based Synaptic Plasticity
cs.NE cs.AI cs.LG
Online synaptic plasticity rules derived from gradient descent achieve high accuracy on a wide range of practical tasks. However, their software implementation often requires tediously hand-derived gradients or using gradient backpropagation which sacrifices the online capability of the rules. In this work, we present a custom automatic differentiation (AD) pipeline for sparse and online implementation of gradient-based synaptic plasticity rules that generalizes to arbitrary neuron models. Our work combines the programming ease of backpropagation-type methods for forward AD while being memory-efficient. To achieve this, we exploit the advantageous compute and memory scaling of online synaptic plasticity by providing an inherently sparse implementation of AD where expensive tensor contractions are replaced with simple element-wise multiplications if the tensors are diagonal. Gradient-based synaptic plasticity rules such as eligibility propagation (e-prop) have exactly this property and thus profit immensely from this feature. We demonstrate the alignment of our gradients with respect to gradient backpropagation on an synthetic task where e-prop gradients are exact, as well as audio speech classification benchmarks. We demonstrate how memory utilization scales with network size without dependence on the sequence length, as expected from forward AD methods.
2501.11409
Unsupervised Learning in Echo State Networks for Input Reconstruction
cs.LG cs.AI eess.SP nlin.CD q-bio.NC
Conventional echo state networks (ESNs) require supervised learning to train the readout layer, using the desired outputs as training data. In this study, we focus on input reconstruction (IR), which refers to training the readout layer to reproduce the input time series in its output. We reformulate the learning algorithm of the ESN readout layer to perform IR using unsupervised learning (UL). By conducting theoretical analysis and numerical experiments, we demonstrate that IR in ESNs can be effectively implemented under realistic conditions without explicitly using the desired outputs as training data; in this way, UL is enabled. Furthermore, we demonstrate that applications relying on IR, such as dynamical system replication and noise filtering, can be reformulated within the UL framework. Our findings establish a theoretically sound and universally applicable IR formulation, along with its related tasks in ESNs. This work paves the way for novel predictions and highlights unresolved theoretical challenges in ESNs, particularly in the context of time-series processing methods and computational models of the brain.
2501.11411
Beyond the Hype: Benchmarking LLM-Evolved Heuristics for Bin Packing
cs.NE
Coupling Large Language Models (LLMs) with Evolutionary Algorithms has recently shown significant promise as a technique to design new heuristics that outperform existing methods, particularly in the field of combinatorial optimisation. An escalating arms race is both rapidly producing new heuristics and improving the efficiency of the processes evolving them. However, driven by the desire to quickly demonstrate the superiority of new approaches, evaluation of the new heuristics produced for a specific domain is often cursory: testing on very few datasets in which instances all belong to a specific class from the domain, and on few instances per class. Taking bin-packing as an example, to the best of our knowledge we conduct the first rigorous benchmarking study of new LLM-generated heuristics, comparing them to well-known existing heuristics across a large suite of benchmark instances using three performance metrics. For each heuristic, we then evolve new instances won by the heuristic and perform an instance space analysis to understand where in the feature space each heuristic performs well. We show that most of the LLM heuristics do not generalise well when evaluated across a broad range of benchmarks in contrast to existing simple heuristics, and suggest that any gains from generating very specialist heuristics that only work in small areas of the instance space need to be weighed carefully against the considerable cost of generating these heuristics.
2501.11413
Generalization and Informativeness of Weighted Conformal Risk Control Under Covariate Shift
cs.LG cs.AI cs.IT math.IT
Predictive models are often required to produce reliable predictions under statistical conditions that are not matched to the training data. A common type of training-testing mismatch is covariate shift, where the conditional distribution of the target variable given the input features remains fixed, while the marginal distribution of the inputs changes. Weighted conformal risk control (W-CRC) uses data collected during the training phase to convert point predictions into prediction sets with valid risk guarantees at test time despite the presence of a covariate shift. However, while W-CRC provides statistical reliability, its efficiency -- measured by the size of the prediction sets -- can only be assessed at test time. In this work, we relate the generalization properties of the base predictor to the efficiency of W-CRC under covariate shifts. Specifically, we derive a bound on the inefficiency of the W-CRC predictor that depends on algorithmic hyperparameters and task-specific quantities available at training time. This bound offers insights on relationships between the informativeness of the prediction sets, the extent of the covariate shift, and the size of the calibration and training sets. Experiments on fingerprinting-based localization validate the theoretical results.
2501.11414
Algorithm Selection with Probing Trajectories: Benchmarking the Choice of Classifier Model
cs.LG cs.NE
Recent approaches to training algorithm selectors in the black-box optimisation domain have advocated for the use of training data that is algorithm-centric in order to encapsulate information about how an algorithm performs on an instance, rather than relying on information derived from features of the instance itself. Probing-trajectories that consist of a sequence of objective performance per function evaluation obtained from a short run of an algorithm have recently shown particular promise in training accurate selectors. However, training models on this type of data requires an appropriately chosen classifier given the sequential nature of the data. There are currently no clear guidelines for choosing the most appropriate classifier for algorithm selection using time-series data from the plethora of models available. To address this, we conduct a large benchmark study using 17 different classifiers and three types of trajectory on a classification task using the BBOB benchmark suite using both leave-one-instance out and leave-one-problem out cross-validation. In contrast to previous studies using tabular data, we find that the choice of classifier has a significant impact, showing that feature-based and interval-based models are the best choices.
2501.11416
Mapping network structures and dynamics of decentralised cryptocurrencies: The evolution of Bitcoin (2009-2023)
cs.CE
Cryptocurrencies have recently been in the spotlight of public debate due to their embrace by the new US President, with crypto fans expecting a 'bull run'. The global cryptocurrency market capitalisation is more than \$3.50 trillion, with 1 Bitcoin exchanging for more than \$97,000 at the end of November 2024. Monitoring the evolution of these systems is key to understanding whether the popular perception of cryptocurrencies as a new, sustainable economic infrastructure is well-founded. In this paper, we have reconstructed the network structures and dynamics of Bitcoin from its launch in January 2009 to December 2023 and identified its key evolutionary phases. Our results show that network centralisation and wealth concentration increased from the very early years, following a richer-get-richer mechanism. This trend was endogenous to the system, beyond any subsequent institutional or exogenous influence. The evolution of Bitcoin is characterised by three periods, Exploration, Adaptation and Maturity, with substantial coherent network patterns. Our findings suggest that Bitcoin is a highly centralised structure, with high levels of wealth inequality and internally crystallised power dynamics, which may have negative implications for its long-term sustainability.
2501.11417
Neural Contextual Reinforcement Framework for Logical Structure Language Generation
cs.CL cs.AI
The Neural Contextual Reinforcement Framework introduces an innovative approach to enhancing the logical coherence and structural consistency of text generated by large language models. Leveraging reinforcement learning principles, the framework integrates custom reward functions and dynamic context alignment mechanisms to address challenges inherent in maintaining long-range dependencies across extended sequences. The architecture incorporates multi-head attention layers and hierarchical encoding modules, enabling the model to produce outputs that align closely with human expectations of logical structure and semantic flow. Quantitative evaluations across diverse datasets demonstrate substantial improvements in coherence metrics, perplexity reduction, and semantic alignment, showcasing the framework's ability to outperform baseline models in both general and domain-specific tasks. Qualitative analyses further highlight the framework's capacity to generate text with improved narrative clarity and reduced redundancy, reflecting its effectiveness in balancing fluency with structural precision. In addition to its performance gains, the framework exhibits robustness in handling noisy input data and scalability across varying model sizes, reinforcing its versatility in practical applications. Experimental results reveal that optimal context window sizes significantly influence coherence outcomes, showing the importance of architectural flexibility in adapting to diverse linguistic structures. Cross-lingual performance evaluations affirm the framework's adaptability to multiple languages, extending its utility beyond monolingual contexts. Resource efficiency analyses indicate a reduction in computational overhead compared to traditional approaches, emphasizing the practicality of the framework for large-scale deployment.
2501.11419
An Analysis of the Correctness and Computational Complexity of Path Planning in Payment Channel Networks
cs.DM cs.CE
Payment Channel Networks (PCNs) are a method for improving the scaling and latency of cryptocurrency transactions. For a payment to be made between two peers in a PCN, a feasible low-fee path in the network must be planned. Many PCN path planning algorithms use a search algorithm that is a variant of Dijkstra's algorithm. In this article, we prove the correctness and computational complexity of this algorithm. Specifically, we show that, if the PCN satisfies a consistency property relating to the fees charged by payment channels, the algorithm is correct and has polynomial computational complexity. However, in the general case, the algorithm is not correct and the path planning problem is NP-hard. These newly developed results can be used to inform the development of new or existing PCNs amenable to path planning. For example, we show that the Lightning Network, which is the most widely used PCN and is built on the Bitcoin cryptocurrency, currently satisfies the above consistency property. As a second contribution, we demonstrate that a small modification to the above path planning algorithm which, although having the same asymptotic computational complexity, empirically shows better performance. This modification involves the use of a bidirectional search and is empirically evaluated by simulating transactions on the Lightning Network.
2501.11421
Online Clustering with Bandit Information
cs.LG cs.IT math.IT math.ST stat.TH
We study the problem of online clustering within the multi-armed bandit framework under the fixed confidence setting. In this multi-armed bandit problem, we have $M$ arms, each providing i.i.d. samples that follow a multivariate Gaussian distribution with an {\em unknown} mean and a known unit covariance. The arms are grouped into $K$ clusters based on the distance between their means using the Single Linkage (SLINK) clustering algorithm on the means of the arms. Since the true means are unknown, the objective is to obtain the above clustering of the arms with the minimum number of samples drawn from the arms, subject to an upper bound on the error probability. We introduce a novel algorithm, Average Tracking Bandit Online Clustering (ATBOC), and prove that this algorithm is order optimal, meaning that the upper bound on its expected sample complexity for given error probability $\delta$ is within a factor of 2 of an instance-dependent lower bound as $\delta \rightarrow 0$. Furthermore, we propose a computationally more efficient algorithm, Lower and Upper Confidence Bound-based Bandit Online Clustering (LUCBBOC), inspired by the LUCB algorithm for best arm identification. Simulation results demonstrate that the performance of LUCBBOC is comparable to that of ATBOC. We numerically assess the effectiveness of the proposed algorithms through numerical experiments on both synthetic datasets and the real-world MovieLens dataset. To the best of our knowledge, this is the first work on bandit online clustering that allows arms with different means in a cluster and $K$ greater than 2.
2501.11422
Multi-View Spectral Clustering for Graphs with Multiple View Structures
cs.LG cs.AI
Despite the fundamental importance of clustering, to this day, much of the relevant research is still based on ambiguous foundations, leading to an unclear understanding of whether or how the various clustering methods are connected with each other. In this work, we provide an additional stepping stone towards resolving such ambiguities by presenting a general clustering framework that subsumes a series of seemingly disparate clustering methods, including various methods belonging to the widely popular spectral clustering framework. In fact, the generality of the proposed framework is additionally capable of shedding light to the largely unexplored area of multi-view graphs where each view may have differently clustered nodes. In turn, we propose GenClus: a method that is simultaneously an instance of this framework and a generalization of spectral clustering, while also being closely related to k-means as well. This results in a principled alternative to the few existing methods studying this special type of multi-view graphs. Then, we conduct in-depth experiments, which demonstrate that GenClus is more computationally efficient than existing methods, while also attaining similar or better clustering performance. Lastly, a qualitative real-world case-study further demonstrates the ability of GenClus to produce meaningful clusterings.
2501.11425
Agent-R: Training Language Model Agents to Reflect via Iterative Self-Training
cs.AI
Large Language Models (LLMs) agents are increasingly pivotal for addressing complex tasks in interactive environments. Existing work mainly focuses on enhancing performance through behavior cloning from stronger experts, yet such approaches often falter in real-world applications, mainly due to the inability to recover from errors. However, step-level critique data is difficult and expensive to collect. Automating and dynamically constructing self-critique datasets is thus crucial to empowering models with intelligent agent capabilities. In this work, we propose an iterative self-training framework, Agent-R, that enables language Agent to Reflect on the fly. Unlike traditional methods that reward or penalize actions based on correctness, Agent-R leverages MCTS to construct training data that recover correct trajectories from erroneous ones. A key challenge of agent reflection lies in the necessity for timely revision rather than waiting until the end of a rollout. To address this, we introduce a model-guided critique construction mechanism: the actor model identifies the first error step (within its current capability) in a failed trajectory. Starting from it, we splice it with the adjacent correct path, which shares the same parent node in the tree. This strategy enables the model to learn reflection based on its current policy, therefore yielding better learning efficiency. To further explore the scalability of this self-improvement paradigm, we investigate iterative refinement of both error correction capabilities and dataset construction. Our findings demonstrate that Agent-R continuously improves the model's ability to recover from errors and enables timely error correction. Experiments on three interactive environments show that Agent-R effectively equips agents to correct erroneous actions while avoiding loops, achieving superior performance compared to baseline methods (+5.59%).
2501.11428
Enhancing Coronary Artery Calcium Scoring via Multi-Organ Segmentation on Non-Contrast Cardiac Computed Tomography
cs.CV cs.AI cs.LG
Despite coronary artery calcium scoring being considered a largely solved problem within the realm of medical artificial intelligence, this paper argues that significant improvements can still be made. By shifting the focus from pathology detection to a deeper understanding of anatomy, the novel algorithm proposed in the paper both achieves high accuracy in coronary artery calcium scoring and offers enhanced interpretability of the results. This approach not only aids in the precise quantification of calcifications in coronary arteries, but also provides valuable insights into the underlying anatomical structures. Through this anatomically-informed methodology, the paper shows how a nuanced understanding of the heart's anatomy can lead to more accurate and interpretable results in the field of cardiovascular health. We demonstrate the superior accuracy of the proposed method by evaluating it on an open-source multi-vendor dataset, where we obtain results at the inter-observer level, surpassing the current state of the art. Finally, the qualitative analyses show the practical value of the algorithm in such tasks as labeling coronary artery calcifications, identifying aortic calcifications, and filtering out false positive detections due to noise.
2501.11429
The Explanation Game -- Rekindled (Extended Version)
cs.AI
Recent work demonstrated the existence of critical flaws in the current use of Shapley values in explainable AI (XAI), i.e. the so-called SHAP scores. These flaws are significant in that the scores provided to a human decision-maker can be misleading. Although these negative results might appear to indicate that Shapley values ought not be used in XAI, this paper argues otherwise. Concretely, this paper proposes a novel definition of SHAP scores that overcomes existing flaws. Furthermore, the paper outlines a practically efficient solution for the rigorous estimation of the novel SHAP scores. Preliminary experimental results confirm our claims, and further underscore the flaws of the current SHAP scores.
2501.11430
A Survey on Diffusion Models for Anomaly Detection
cs.LG cs.AI
Diffusion models (DMs) have emerged as a powerful class of generative AI models, showing remarkable potential in anomaly detection (AD) tasks across various domains, such as cybersecurity, fraud detection, healthcare, and manufacturing. The intersection of these two fields, termed diffusion models for anomaly detection (DMAD), offers promising solutions for identifying deviations in increasingly complex and high-dimensional data. In this survey, we review recent advances in DMAD research. We begin by presenting the fundamental concepts of AD and DMs, followed by a comprehensive analysis of classic DM architectures including DDPMs, DDIMs, and Score SDEs. We further categorize existing DMAD methods into reconstruction-based, density-based, and hybrid approaches, providing detailed examinations of their methodological innovations. We also explore the diverse tasks across different data modalities, encompassing image, time series, video, and multimodal data analysis. Furthermore, we discuss critical challenges and emerging research directions, including computational efficiency, model interpretability, robustness enhancement, edge-cloud collaboration, and integration with large language models. The collection of DMAD research papers and resources is available at https://github.com/fdjingliu/DMAD.
2501.11434
An Incremental Sampling and Segmentation-Based Approach for Motion Planning Infeasibility
cs.RO
We present a simple and easy-to-implement algorithm to detect plan infeasibility in kinematic motion planning. Our method involves approximating the robot's configuration space to a discrete space, where each degree of freedom has a finite set of values. The obstacle region separates the free configuration space into different connected regions. For a path to exist between the start and goal configurations, they must lie in the same connected region of the free space. Thus, to ascertain plan infeasibility, we merely need to sample adequate points from the obstacle region that isolate start and goal. Accordingly, we progressively construct the configuration space by sampling from the discretized space and updating the bitmap cells representing obstacle regions. Subsequently, we partition this partially built configuration space to identify different connected components within it and assess the connectivity of the start and goal cells. We illustrate this methodology on five different scenarios with configuration spaces having up to 5 degree-of-freedom (DOF).
2501.11440
RACCOON: A Retrieval-Augmented Generation Approach for Location Coordinate Capture from News Articles
cs.CL
Geocoding involves automatic extraction of location coordinates of incidents reported in news articles, and can be used for epidemic intelligence or disaster management. This paper introduces Retrieval-Augmented Coordinate Capture Of Online News articles (RACCOON), an open-source geocoding approach that extracts geolocations from news articles. RACCOON uses a retrieval-augmented generation (RAG) approach where candidate locations and associated information are retrieved in the form of context from a location database, and a prompt containing the retrieved context, location mentions and news articles is fed to an LLM to generate the location coordinates. Our evaluation on three datasets, two underlying LLMs, three baselines and several ablation tests based on the components of RACCOON demonstrate the utility of RACCOON. To the best of our knowledge, RACCOON is the first RAG-based approach for geocoding using pre-trained LLMs.
2501.11441
Ontology Matching with Large Language Models and Prioritized Depth-First Search
cs.IR cs.CL
Ontology matching (OM) plays a key role in enabling data interoperability and knowledge sharing, but it remains challenging due to the need for large training datasets and limited vocabulary processing in machine learning approaches. Recently, methods based on Large Language Model (LLMs) have shown great promise in OM, particularly through the use of a retrieve-then-prompt pipeline. In this approach, relevant target entities are first retrieved and then used to prompt the LLM to predict the final matches. Despite their potential, these systems still present limited performance and high computational overhead. To address these issues, we introduce MILA, a novel approach that embeds a retrieve-identify-prompt pipeline within a prioritized depth-first search (PDFS) strategy. This approach efficiently identifies a large number of semantic correspondences with high accuracy, limiting LLM requests to only the most borderline cases. We evaluated MILA using the biomedical challenge proposed in the 2023 and 2024 editions of the Ontology Alignment Evaluation Initiative. Our method achieved the highest F-Measure in four of the five unsupervised tasks, outperforming state-of-the-art OM systems by up to 17%. It also performed better than or comparable to the leading supervised OM systems. MILA further exhibited task-agnostic performance, remaining stable across all tasks and settings, while significantly reducing LLM requests. These findings highlight that high-performance LLM-based OM can be achieved through a combination of programmed (PDFS), learned (embedding vectors), and prompting-based heuristics, without the need of domain-specific heuristics or fine-tuning.
2501.11447
Decomposing Interventional Causality into Synergistic, Redundant, and Unique Components
cs.AI cs.IT math.IT physics.data-an
We introduce a novel framework for decomposing interventional causal effects into synergistic, redundant, and unique components, building on the intuition of Partial Information Decomposition (PID) and the principle of M\"obius inversion. While recent work has explored a similar decomposition of an observational measure, we argue that a proper causal decomposition must be interventional in nature. We develop a mathematical approach that systematically quantifies how causal power is distributed among variables in a system, using a recently derived closed-form expression for the M\"obius function of the redundancy lattice. The formalism is then illustrated by decomposing the causal power in logic gates, cellular automata, and chemical reaction networks. Our results reveal how the distribution of causal power can be context- and parameter-dependent. This decomposition provides new insights into complex systems by revealing how causal influences are shared and combined among multiple variables, with potential applications ranging from attribution of responsibility in legal or AI systems, to the analysis of biological networks or climate models.
2501.11453
Integrate-and-Fire from a Mathematical and Signal Processing Perspective
eess.SP cs.NE
Integrate-and-Fire (IF) is an idealized model of the spike-triggering mechanism of a biological neuron. It is used to realize the bio-inspired event-based principle of information processing in neuromorphic computing. We show that IF is closely related to the concept of Send-on-Delta (SOD) as used in threshold-based sampling. It turns out that the IF model can be adjusted in a way that SOD can be understood as differential version of IF. As a result, we gain insight into the underlying metric structure based on the Alexiewicz norm with consequences for clarifying the underlying signal space including bounded integrable signals with superpositions of finitely many Dirac impulses, the identification of a maximum sparsity property, error bounds for signal reconstruction and a characterization in terms of sparse regularization.
2501.11454
Improving thermal state preparation of Sachdev-Ye-Kitaev model with reinforcement learning on quantum hardware
quant-ph cs.AI cs.LG hep-lat hep-th
The Sachdev-Ye-Kitaev (SYK) model, known for its strong quantum correlations and chaotic behavior, serves as a key platform for quantum gravity studies. However, variationally preparing thermal states on near-term quantum processors for large systems (N>12, where N is the number of Majorana fermions) presents a significant challenge due to the rapid growth in the complexity of parameterized quantum circuits. This paper addresses this challenge by integrating reinforcement learning (RL) with convolutional neural networks, employing an iterative approach to optimize the quantum circuit and its parameters. The refinement process is guided by a composite reward signal derived from entropy and the expectation values of the SYK Hamiltonian. This approach reduces the number of CNOT gates by two orders of magnitude for systems N>10 compared to traditional methods like first-order Trotterization. We demonstrate the effectiveness of the RL framework in both noiseless and noisy quantum hardware environments, maintaining high accuracy in thermal state preparation. This work contributes to the advancement of a scalable, RL-based framework with applications for computations of thermal out-of-time-order correlators in quantum many-body systems and quantum gravity studies on near-term quantum hardware.
2501.11459
Multi-Stage Active Sequential Hypothesis Testing with Clustered Hypotheses
cs.IT math.IT
We consider the problem where an active Decision-Maker (DM) is tasked to identify the true hypothesis using as few as possible observations while maintaining accuracy. The DM collects observations according to its determined actions and knows the distributions under each hypothesis. We propose a deterministic and adaptive multi-stage hypothesis-elimination strategy where the DM selects an action, applies it repeatedly, and discards hypotheses in light of its obtained observations. The DM selects actions based on maximal separation expressed by the distance between the parameter vectors of each distribution under each hypothesis. Close distributions can be clustered, simplifying the search and significantly reducing the number of required observations. Our algorithms achieve vanishing Average Bayes Risk (ABR) as the error probability approaches zero, i.e., the algorithm is asymptotically optimal. Furthermore, we show that the ABR is bounded when the number of hypotheses grows. Simulations are carried out to evaluate the algorithm's performance compared to another multi-stage hypothesis-elimination algorithm, where an improvement of several orders of magnitude in the mean number of observations required is observed.
2501.11462
On the Adversarial Vulnerabilities of Transfer Learning in Remote Sensing
cs.CV eess.IV
The use of pretrained models from general computer vision tasks is widespread in remote sensing, significantly reducing training costs and improving performance. However, this practice also introduces vulnerabilities to downstream tasks, where publicly available pretrained models can be used as a proxy to compromise downstream models. This paper presents a novel Adversarial Neuron Manipulation method, which generates transferable perturbations by selectively manipulating single or multiple neurons in pretrained models. Unlike existing attacks, this method eliminates the need for domain-specific information, making it more broadly applicable and efficient. By targeting multiple fragile neurons, the perturbations achieve superior attack performance, revealing critical vulnerabilities in deep learning models. Experiments on diverse models and remote sensing datasets validate the effectiveness of the proposed method. This low-access adversarial neuron manipulation technique highlights a significant security risk in transfer learning models, emphasizing the urgent need for more robust defenses in their design when addressing the safety-critical remote sensing tasks.
2501.11463
Curiosity-Driven Reinforcement Learning from Human Feedback
cs.CL
Reinforcement learning from human feedback (RLHF) has proven effective in aligning large language models (LLMs) with human preferences, but often at the cost of reduced output diversity. This trade-off between diversity and alignment quality remains a significant challenge. Drawing inspiration from curiosity-driven exploration in reinforcement learning, we introduce curiosity-driven RLHF (CD-RLHF), a framework that incorporates intrinsic rewards for novel states, alongside traditional sparse extrinsic rewards, to optimize both output diversity and alignment quality. We demonstrate the effectiveness of CD-RLHF through extensive experiments on a range of tasks, including text summarization and instruction following. Our approach achieves significant gains in diversity on multiple diversity-oriented metrics while maintaining alignment with human preferences comparable to standard RLHF. We make our code publicly available at https://github.com/ernie-research/CD-RLHF.
2501.11467
Fixed Point Certificates for Reachability and Expected Rewards in MDPs
cs.LO cs.DM cs.SY eess.SY
The possibility of errors in human-engineered formal verification software, such as model checkers, poses a serious threat to the purpose of these tools. An established approach to mitigate this problem are certificates -- lightweight, easy-to-check proofs of the verification results. In this paper, we develop novel certificates for model checking of Markov decision processes (MDPs) with quantitative reachability and expected reward properties. Our approach is conceptually simple and relies almost exclusively on elementary fixed point theory. Our certificates work for arbitrary finite MDPs and can be readily computed with little overhead using standard algorithms. We formalize the soundness of our certificates in Isabelle/HOL and provide a formally verified certificate checker. Moreover, we augment existing algorithms in the probabilistic model checker Storm with the ability to produce certificates and demonstrate practical applicability by conducting the first formal certification of the reference results in the Quantitative Verification Benchmark Set.
2501.11469
MASS: Overcoming Language Bias in Image-Text Matching
cs.CV cs.LG
Pretrained visual-language models have made significant advancements in multimodal tasks, including image-text retrieval. However, a major challenge in image-text matching lies in language bias, where models predominantly rely on language priors and neglect to adequately consider the visual content. We thus present Multimodal ASsociation Score (MASS), a framework that reduces the reliance on language priors for better visual accuracy in image-text matching problems. It can be seamlessly incorporated into existing visual-language models without necessitating additional training. Our experiments have shown that MASS effectively lessens language bias without losing an understanding of linguistic compositionality. Overall, MASS offers a promising solution for enhancing image-text matching performance in visual-language models.
2501.11473
Strong Data Processing Properties of R\'enyi-divergences via Pinsker-type Inequalities
cs.IT math.IT
We investigate strong data processing inequalities (SDPIs) of the R\'enyi-divergence between two discrete distributions when both distributions are passed through a fixed channel. We provide a condition on the channel for which the DPI holds with equality given two arbitrary distributions in the probability simplex. Motivated by this, we examine the contraction behavior for restricted sets of prior distributions via $f$-divergence inequalities: We provide an alternative proof of the optimal reverse Pinsker's inequality for R\'enyi-divergences first shown by Binette. We further present an improved Pinsker's inequality for R\'enyi-divergence based on the joint range technique by Harremo\"es and Vajda. The presented bound is tight whenever the value of the total variation distance is larger than $\frac{1}{\alpha}$. By framing these inequalities in a cross-channel setting, we arrive at SDPIs that can be adapted to use-case specific restrictions of input distribution and channel. We apply these results to the R\'enyi local differential privacy amplification through post-processing by channels that satisfy no local differential privacy guarantee.
2501.11477
QGAIC: Quantum Inspired Genetic Algorithm for Image Classification
cs.NE
This study uses two meta-heuristics methodologies to introduce two novel quantum-inspired meta heuristic approaches: quantum-inspired genetic algorithm (QIGA1) and quantum-inspired genetic algorithm with dynamic approach (QIGA2). The two suggested methods combine a classical and quantum genetic algorithm approach. Both approaches use The correlation coefficient as an assessment function to identify the best (optimal) values for binary image. In quantum computing, they use simple ideas like qubits and state superposition. Due to these characteristics, parallelism which uses the time discreteness of quantum mechanical systems, is exhibited. For five distinct MNIST datasets, the performance of all participating algorithms has been assessed by comparing the suggested approaches first with their traditional approach counterparts and then with the proposed methods QIGA1 and QIGA2. Each method's ideal threshold value, associated fitness value (best and average), loss, and accuracy for each MNIST dataset have all been published. The outcomes demonstrate the superior efficiency of the suggested approaches over their traditional equivalents.
2501.11478
Each Graph is a New Language: Graph Learning with LLMs
cs.CL cs.AI cs.LG
Recent efforts leverage Large Language Models (LLMs) for modeling text-attributed graph structures in node classification tasks. These approaches describe graph structures for LLMs to understand or aggregate LLM-generated textual attribute embeddings through graph structure. However, these approaches face two main limitations in modeling graph structures with LLMs. (i) Graph descriptions become verbose in describing high-order graph structure. (ii) Textual attributes alone do not contain adequate graph structure information. It is challenging to model graph structure concisely and adequately with LLMs. LLMs lack built-in mechanisms to model graph structures directly. They also struggle with complex long-range dependencies between high-order nodes and target nodes. Inspired by the observation that LLMs pre-trained on one language can achieve exceptional performance on another with minimal additional training, we propose \textbf{G}raph-\textbf{D}efined \textbf{L}anguage for \textbf{L}arge \textbf{L}anguage \textbf{M}odel (GDL4LLM). This novel framework enables LLMs to transfer their powerful language understanding capabilities to graph-structured data. GDL4LLM translates graphs into a graph language corpus instead of graph descriptions and pre-trains LLMs on this corpus to adequately understand graph structures. During fine-tuning, this corpus describes the structural information of target nodes concisely with only a few tokens. By treating graphs as a new language, GDL4LLM enables LLMs to model graph structures adequately and concisely for node classification tasks. Extensive experiments on three real-world datasets demonstrate that GDL4LLM outperforms description-based and textual attribute embeddings-based baselines by efficiently modeling different orders of graph structure with LLMs.
2501.11485
SimLabel: Consistency-Guided OOD Detection with Pretrained Vision-Language Models
cs.CV
Detecting out-of-distribution (OOD) data is crucial in real-world machine learning applications, particularly in safety-critical domains. Existing methods often leverage language information from vision-language models (VLMs) to enhance OOD detection by improving confidence estimation through rich class-wise text information. However, when building OOD detection score upon on in-distribution (ID) text-image affinity, existing works either focus on each ID class or whole ID label sets, overlooking inherent ID classes' connection. We find that the semantic information across different ID classes is beneficial for effective OOD detection. We thus investigate the ability of image-text comprehension among different semantic-related ID labels in VLMs and propose a novel post-hoc strategy called SimLabel. SimLabel enhances the separability between ID and OOD samples by establishing a more robust image-class similarity metric that considers consistency over a set of similar class labels. Extensive experiments demonstrate the superior performance of SimLabel on various zero-shot OOD detection benchmarks. The proposed model is also extended to various VLM-backbones, demonstrating its good generalization ability. Our demonstration and implementation codes are available at: https://github.com/ShuZou-1/SimLabel.
2501.11487
Detecting Convolutional Codes: A Markovian Approach with LRT and DNN
cs.IT math.IT
Identifying the unknown convolutional code corresponding to the given intercepted data is an important problem in military surveillance and in wireless communication. While a variety of code identification algorithms are available in the literature, the key contribution of our work lies in the novel solution and the corresponding analysis. In this paper, we focus on the situation when the given data corresponds to either of the two potential convolutional codes and the goal is to detect the correct code. We first provide a new interpretation of the convolutional code as a Markov chain, which is more suitable for analyzing the code detection problem. Our problem then gets reduced to identifying between the two Markov chains. We provide the closed-form expressions for the corresponding state transition matrices and estimate the error exponent for the underlying likelihood ratio test (LRT). We also provide a computationally efficient BCJR-based method for computing the likelihoods required for the LRT. We observe that BCJR-based likelihoods suffer from numerical issues for a longer data sequence, and hence, in this case, we design neural networks that have been found to achieve the optimal performance of the LRT.
2501.11493
Communication-Efficient Federated Learning Based on Explanation-Guided Pruning for Remote Sensing Image Classification
cs.CV cs.AI
Federated learning (FL) is a decentralized machine learning paradigm, where multiple clients collaboratively train a global model by exchanging only model updates with the central server without sharing the local data of clients. Due to the large volume of model updates required to be transmitted between clients and the central server, most FL systems are associated with high transfer costs (i.e., communication overhead). This issue is more critical for operational applications in remote sensing (RS), especially when large-scale RS data is processed and analyzed through FL systems with restricted communication bandwidth. To address this issue, we introduce an explanation-guided pruning strategy for communication-efficient FL in the context of RS image classification. Our pruning strategy is defined based on the layerwise relevance propagation (LRP) driven explanations to: 1) efficiently and effectively identify the most relevant and informative model parameters (to be exchanged between clients and the central server); and 2) eliminate the non-informative ones to minimize the volume of model updates. The experimental results on the BigEarthNet-S2 dataset demonstrate that our strategy effectively reduces the number of shared model updates, while increasing the generalization ability of the global model. The code of this work will be publicly available at https://git.tu-berlin.de/rsim/FL-LRP
2501.11495
Discrete-Time Passivity-Based Control using Hermite-Obreschkoff Methods
eess.SY cs.SY
The motivation for this paper is the implementation of nonlinear state feedback control, designed based on the continuous-time plant model, in a sampled control loop under relatively slow sampling. In previous work we have shown that using one-step predictions of the target dynamics with higher order integration schemes, together with possibly higher order input shaping, is a simple and effective way to increase the feasible sampling times until performance degradation and instability occur. In this contribution we present a unifying derivation for arbitrary orders of the previously used Lobatto IIIA collocation and Hermite interpolation schemes through the Hermite-Obreschkoff formula. We derive, moreover, an IDA-PBC controller for a magnetic levitation system, which requires a non-constant target interconnection matrix, and show experimental results.
2501.11496
Generative AI and Large Language Models in Language Preservation: Opportunities and Challenges
cs.CL cs.AI cs.LG
Generative AI and large-scale language models (LLM) have emerged as powerful tools in language preservation, particularly for near-native and endangered languages. With the increasing reliance on technology for communication, education, and cultural documentation, new opportunities have emerged to mitigate the dramatic decline of linguistic diversity worldwide. This paper examines the role of generative AIs and LLMs in preserving endangered languages, highlighting the risks and challenges associated with their use. We analyze the underlying technologies driving these models, including natural language processing (NLP) and deep learning, and explore several cases where these technologies have been applied to low-resource languages. Additionally, we discuss ethical considerations, data scarcity issues, and technical challenges while proposing solutions to enhance AI-driven language preservation.
2501.11498
Dialect2SQL: A Novel Text-to-SQL Dataset for Arabic Dialects with a Focus on Moroccan Darija
cs.SE cs.AI cs.CL cs.DB
The task of converting natural language questions (NLQs) into executable SQL queries, known as text-to-SQL, has gained significant interest in recent years, as it enables non-technical users to interact with relational databases. Many benchmarks, such as SPIDER and WikiSQL, have contributed to the development of new models and the evaluation of their performance. In addition, other datasets, like SEDE and BIRD, have introduced more challenges and complexities to better map real-world scenarios. However, these datasets primarily focus on high-resource languages such as English and Chinese. In this work, we introduce Dialect2SQL, the first large-scale, cross-domain text-to-SQL dataset in an Arabic dialect. It consists of 9,428 NLQ-SQL pairs across 69 databases in various domains. Along with SQL-related challenges such as long schemas, dirty values, and complex queries, our dataset also incorporates the complexities of the Moroccan dialect, which is known for its diverse source languages, numerous borrowed words, and unique expressions. This demonstrates that our dataset will be a valuable contribution to both the text-to-SQL community and the development of resources for low-resource languages.
2501.11499
KEIR @ ECIR 2025: The Second Workshop on Knowledge-Enhanced Information Retrieval
cs.IR
Pretrained language models (PLMs) like BERT and GPT-4 have become the foundation for modern information retrieval (IR) systems. However, existing PLM-based IR models primarily rely on the knowledge learned during training for prediction, limiting their ability to access and incorporate external, up-to-date, or domain-specific information. Therefore, current information retrieval systems struggle with semantic nuances, context relevance, and domain-specific issues. To address these challenges, we propose the second Knowledge-Enhanced Information Retrieval workshop (KEIR @ ECIR 2025) as a platform to discuss innovative approaches that integrate external knowledge, aiming to enhance the effectiveness of information retrieval in a rapidly evolving technological landscape. The goal of this workshop is to bring together researchers from academia and industry to discuss various aspects of knowledge-enhanced information retrieval.
2501.11502
Hierarchical Coded Caching in High Memory Regime with Coded Placement
cs.IT math.IT
We consider a two-layer hierarchical coded caching network where a server with a library of $N$ files is connected to $K_1$ mirrors, each having a cache memory of size $M_1$. Each mirror is further connected to $K_2$ users, each equipped with a dedicated cache of size $M_2$. In this paper, we propose two distinct coded caching schemes based on coded placement, corresponding to two distinct memory pairs, \( (M_1, M_2) \). We show that the proposed schemes outperform the existing schemes at these memory points given by the proposed schemes for smaller values of $K_2$. In setups where mirrors are positioned near each other, avoiding signal interference is crucial. This can be ensured by having all mirrors transmit using orthogonal carrier frequencies. To compare our schemes with existing ones, we used the composite rate metric, which accurately represents the total bandwidth utilized in such setups. The composite rate is given by $\overline{R} = R_1 + K_1 R_2$, where $R_1$ is the rate from the server to the mirrors, and $R_2$ is the rate from the mirrors to the users, with respect to $M_1$ and $M_2$.
2501.11505
Sun-Jafar-Type Schemes for Weak Private Information Retrieval
cs.IT math.IT
In information-theoretic private information retrieval (PIR), a client wants to retrieve one desired file out of $M$ files, stored across $N$ servers, while keeping the index of the desired file private from each $T$-sized subset of servers. A PIR protocol must ideally maximize the rate, which is the ratio of the file size to the total quantum of the download from the servers, while ensuring such privacy. In Weak-PIR (WPIR), the criterion of perfect information-theoretic privacy is relaxed. This enables higher rates to be achieved, while some information about the desired file index leaks to the servers. This leakage is captured by various known privacy metrics. By leveraging the well-established capacity-achieving schemes of Sun and Jafar under non-colluding ($T=1$) and colluding ($1<T\leq N$) scenarios, we present WPIR protocols for these scenarios. We also present a new WPIR scheme for the MDS scenario, by building upon the scheme by Banawan and Ulukus for this scenario. We present corresponding explicit rate-privacy trade-offs for these setups, under the mutual-information and the maximal leakage privacy metrics. In the collusion-free setup, our presented rate-privacy trade-off under maximal leakage matches that of the previous state of the art. With respect to the MDS scenario under the maximal leakage metric, we compare with the non-explicit trade-off in the literature, and show that our scheme performs better for some numerical examples. For the $T$-collusion setup (under both privacy metrics) and for the MDS setup under the mutual information metric, our rate-privacy trade-offs are the first in the literature, to the best of our knowledge.
2501.11508
See In Detail: Enhancing Sparse-view 3D Gaussian Splatting with Local Depth and Semantic Regularization
cs.CV
3D Gaussian Splatting (3DGS) has shown remarkable performance in novel view synthesis. However, its rendering quality deteriorates with sparse inphut views, leading to distorted content and reduced details. This limitation hinders its practical application. To address this issue, we propose a sparse-view 3DGS method. Given the inherently ill-posed nature of sparse-view rendering, incorporating prior information is crucial. We propose a semantic regularization technique, using features extracted from the pretrained DINO-ViT model, to ensure multi-view semantic consistency. Additionally, we propose local depth regularization, which constrains depth values to improve generalization on unseen views. Our method outperforms state-of-the-art novel view synthesis approaches, achieving up to 0.4dB improvement in terms of PSNR on the LLFF dataset, with reduced distortion and enhanced visual quality.
2501.11511
Subjective and Objective Quality Assessment of Non-Uniformly Distorted Omnidirectional Images
eess.IV cs.CV
Omnidirectional image quality assessment (OIQA) has been one of the hot topics in IQA with the continuous development of VR techniques, and achieved much success in the past few years. However, most studies devote themselves to the uniform distortion issue, i.e., all regions of an omnidirectional image are perturbed by the ``same amount'' of noise, while ignoring the non-uniform distortion issue, i.e., partial regions undergo ``different amount'' of perturbation with the other regions in the same omnidirectional image. Additionally, nearly all OIQA models are verified on the platforms containing a limited number of samples, which largely increases the over-fitting risk and therefore impedes the development of OIQA. To alleviate these issues, we elaborately explore this topic from both subjective and objective perspectives. Specifically, we construct a large OIQA database containing 10,320 non-uniformly distorted omnidirectional images, each of which is generated by considering quality impairments on one or two camera len(s). Then we meticulously conduct psychophysical experiments and delve into the influence of both holistic and individual factors (i.e., distortion range and viewing condition) on omnidirectional image quality. Furthermore, we propose a perception-guided OIQA model for non-uniform distortion by adaptively simulating users' viewing behavior. Experimental results demonstrate that the proposed model outperforms state-of-the-art methods. The source code is available at https://github.com/RJL2000/OIQAND.
2501.11512
Multitask Auxiliary Network for Perceptual Quality Assessment of Non-Uniformly Distorted Omnidirectional Images
eess.IV cs.CV
Omnidirectional image quality assessment (OIQA) has been widely investigated in the past few years and achieved much success. However, most of existing studies are dedicated to solve the uniform distortion problem in OIQA, which has a natural gap with the non-uniform distortion problem, and their ability in capturing non-uniform distortion is far from satisfactory. To narrow this gap, in this paper, we propose a multitask auxiliary network for non-uniformly distorted omnidirectional images, where the parameters are optimized by jointly training the main task and other auxiliary tasks. The proposed network mainly consists of three parts: a backbone for extracting multiscale features from the viewport sequence, a multitask feature selection module for dynamically allocating specific features to different tasks, and auxiliary sub-networks for guiding the proposed model to capture local distortion and global quality change. Extensive experiments conducted on two large-scale OIQA databases demonstrate that the proposed model outperforms other state-of-the-art OIQA metrics, and these auxiliary sub-networks contribute to improve the performance of the proposed model. The source code is available at https://github.com/RJL2000/MTAOIQA.
2501.11513
Transferability of labels between multilens cameras
cs.CV
In this work, a new method for automatically extending Bounding Box (BB) and mask labels across different channels on multilens cameras is presented. For that purpose, the proposed method combines the well known phase correlation method with a refinement process. During the first step, images are aligned by localizing the peak of intensity obtained in the spatial domain after performing the cross correlation process in the frequency domain. The second step consists of obtaining the best possible transformation by using an iterative process maximising the IoU (Intersection over Union) metric. Results show that, by using this method, labels could be transferred across different lens on a camera with an accuracy over 90% in most cases and just by using 65 ms in the whole process. Once the transformations are obtained, artificial RGB images are generated, for labeling them so as to transfer this information into each of the other lens. This work will allow users to use this type of cameras in more fields rather than satellite or medical imagery, giving the chance of labeling even invisible objects in the visible spectrum.
2501.11515
UltraFusion: Ultra High Dynamic Imaging using Exposure Fusion
cs.CV
Capturing high dynamic range (HDR) scenes is one of the most important issues in camera design. Majority of cameras use exposure fusion technique, which fuses images captured by different exposure levels, to increase dynamic range. However, this approach can only handle images with limited exposure difference, normally 3-4 stops. When applying to very high dynamic scenes where a large exposure difference is required, this approach often fails due to incorrect alignment or inconsistent lighting between inputs, or tone mapping artifacts. In this work, we propose UltraFusion, the first exposure fusion technique that can merge input with 9 stops differences. The key idea is that we model the exposure fusion as a guided inpainting problem, where the under-exposed image is used as a guidance to fill the missing information of over-exposed highlight in the over-exposed region. Using under-exposed image as a soft guidance, instead of a hard constrain, our model is robust to potential alignment issue or lighting variations. Moreover, utilizing the image prior of the generative model, our model also generates natural tone mapping, even for very high-dynamic range scene. Our approach outperforms HDR-Transformer on latest HDR benchmarks. Moreover, to test its performance in ultra high dynamic range scene, we capture a new real-world exposure fusion benchmark, UltraFusion Dataset, with exposure difference up to 9 stops, and experiments show that \model~can generate beautiful and high-quality fusion results under various scenarios. An online demo is provided at https://openimaginglab.github.io/UltraFusion/.
2501.11520
Fundus Image Quality Assessment and Enhancement: a Systematic Review
eess.IV cs.CV
As an affordable and convenient eye scan, fundus photography holds the potential for preventing vision impairment, especially in resource-limited regions. However, fundus image degradation is common under intricate imaging environments, impacting following diagnosis and treatment. Consequently, image quality assessment (IQA) and enhancement (IQE) are essential for ensuring the clinical value and reliability of fundus images. While existing reviews offer some overview of this field, a comprehensive analysis of the interplay between IQA and IQE, along with their clinical deployment challenges, is lacking. This paper addresses this gap by providing a thorough review of fundus IQA and IQE algorithms, research advancements, and practical applications. We outline the fundamentals of the fundus photography imaging system and the associated interferences, and then systematically summarize the paradigms in fundus IQA and IQE. Furthermore, we discuss the practical challenges and solutions in deploying IQA and IQE, as well as offer insights into potential future research directions.
2501.11522
Optimal Trajectory Control of Geometrically Exact Strings with Space-Time Finite Elements
eess.SY cs.SY
In this contribution, we present a variational space-time formulation which generates an optimal feed-forward controller for geometrically exact strings. More concretely, the optimization problem is solved with an indirect approach, and the space-time finite element method translates the problem to a set of algebraic equations. Thereby, only the positional field and the corresponding adjoint variable field are approximated by continuous shape functions, which makes the discretization of a velocity field unnecessary. In addition, the variational formulation can be solved using commercial or open source finite element packages. The entire approach can also be interpreted as a multiple-shooting method for solving the optimality conditions based on the semi-discrete problem. The performance of our approach is demonstrated by a numerical test.
2501.11525
Technical Report for the Forgotten-by-Design Project: Targeted Obfuscation for Machine Learning
cs.LG cs.AI cs.CR
The right to privacy, enshrined in various human rights declarations, faces new challenges in the age of artificial intelligence (AI). This paper explores the concept of the Right to be Forgotten (RTBF) within AI systems, contrasting it with traditional data erasure methods. We introduce Forgotten by Design, a proactive approach to privacy preservation that integrates instance-specific obfuscation techniques during the AI model training process. Unlike machine unlearning, which modifies models post-training, our method prevents sensitive data from being embedded in the first place. Using the LIRA membership inference attack, we identify vulnerable data points and propose defenses that combine additive gradient noise and weighting schemes. Our experiments on the CIFAR-10 dataset demonstrate that our techniques reduce privacy risks by at least an order of magnitude while maintaining model accuracy (at 95% significance). Additionally, we present visualization methods for the privacy-utility trade-off, providing a clear framework for balancing privacy risk and model accuracy. This work contributes to the development of privacy-preserving AI systems that align with human cognitive processes of motivated forgetting, offering a robust framework for safeguarding sensitive information and ensuring compliance with privacy regulations.
2501.11526
Meta-Instance Selection. Instance Selection as a Classification Problem with Meta-Features
cs.LG cs.AI
Data pruning, or instance selection, is an important problem in machine learning especially in terms of nearest neighbour classifier. However, in data pruning which speeds up the prediction phase, there is an issue related to the speed and efficiency of the process itself. In response, the study proposes an approach involving transforming the instance selection process into a classification task conducted in a unified meta-feature space where each instance can be classified and assigned to either the "to keep" or "to remove" class. This approach requires training an appropriate meta-classifier, which can be developed based on historical instance selection results from other datasets using reference instance selection methods as a labeling tool. This work proposes constructing the meta-feature space based on properties extracted from the nearest neighbor graph. Experiments conducted on 17 datasets of varying sizes and five reference instance selection methods (ENN, Drop3, ICF, HMN-EI, and CCIS) demonstrate that the proposed solution achieves results comparable to reference instance selection methods while significantly reducing computational complexity. In the proposed approach, the computational complexity of the system depends only on identifying the k-nearest neighbors for each data sample and running the meta-classifier. Additionally, the study discusses the choice of meta-classifier, recommending the use of Balanced Random Forest.
2501.11532
Early Stopping Bayesian Optimization for Controller Tuning
eess.SY cs.SY
Manual tuning of performance-critical controller parameters can be tedious and sub-optimal. Bayesian Optimization (BO) is an increasingly popular practical alternative to automatically optimize controller parameters from few experiments. Standard BO practice is to evaluate the closed-loop performance of parameters proposed during optimization on an episode with a fixed length. However, fixed-length episodes can be wasteful. For example, continuing an episode where already the start shows undesirable behavior such as strong oscillations seems pointless. Therefore, we propose a BO method that stops an episode early if suboptimality becomes apparent before an episode is completed. Such early stopping results in partial observations of the controller's performance, which cannot directly be included in standard BO. We propose three heuristics to facilitate partially observed episodes in BO. Through five numerical and one hardware experiment, we demonstrate that early stopping BO can substantially reduce the time needed for optimization.
2501.11533
The impact of intrinsic rewards on exploration in Reinforcement Learning
cs.AI cs.LG
One of the open challenges in Reinforcement Learning is the hard exploration problem in sparse reward environments. Various types of intrinsic rewards have been proposed to address this challenge by pushing towards diversity. This diversity might be imposed at different levels, favouring the agent to explore different states, policies or behaviours (State, Policy and Skill level diversity, respectively). However, the impact of diversity on the agent's behaviour remains unclear. In this work, we aim to fill this gap by studying the effect of different levels of diversity imposed by intrinsic rewards on the exploration patterns of RL agents. We select four intrinsic rewards (State Count, Intrinsic Curiosity Module (ICM), Maximum Entropy, and Diversity is all you need (DIAYN)), each pushing for a different diversity level. We conduct an empirical study on MiniGrid environment to compare their impact on exploration considering various metrics related to the agent's exploration, namely: episodic return, observation coverage, agent's position coverage, policy entropy, and timeframes to reach the sparse reward. The main outcome of the study is that State Count leads to the best exploration performance in the case of low-dimensional observations. However, in the case of RGB observations, the performance of State Count is highly degraded mostly due to representation learning challenges. Conversely, Maximum Entropy is less impacted, resulting in a more robust exploration, despite being not always optimal. Lastly, our empirical study revealed that learning diverse skills with DIAYN, often linked to improved robustness and generalisation, does not promote exploration in MiniGrid environments. This is because: i) learning the skill space itself can be challenging, and ii) exploration within the skill space prioritises differentiating between behaviours rather than achieving uniform state visitation.
2501.11535
A baseline for machine-learning-based hepatocellular carcinoma diagnosis using multi-modal clinical data
cs.CV
The objective of this paper is to provide a baseline for performing multi-modal data classification on a novel open multimodal dataset of hepatocellular carcinoma (HCC), which includes both image data (contrast-enhanced CT and MRI images) and tabular data (the clinical laboratory test data as well as case report forms). TNM staging is the classification task. Features from the vectorized preprocessed tabular data and radiomics features from contrast-enhanced CT and MRI images are collected. Feature selection is performed based on mutual information. An XGBoost classifier predicts the TNM staging and it shows a prediction accuracy of $0.89 \pm 0.05$ and an AUC of $0.93 \pm 0.03$. The classifier shows that this high level of prediction accuracy can only be obtained by combining image and clinical laboratory data and therefore is a good example case where multi-model classification is mandatory to achieve accurate results.
2501.11538
DenoMAE: A Multimodal Autoencoder for Denoising Modulation Signals
cs.LG
We propose Denoising Masked Autoencoder (Deno-MAE), a novel multimodal autoencoder framework for denoising modulation signals during pretraining. DenoMAE extends the concept of masked autoencoders by incorporating multiple input modalities, including noise as an explicit modality, to enhance cross-modal learning and improve denoising performance. The network is pre-trained using unlabeled noisy modulation signals and constellation diagrams, effectively learning to reconstruct their equivalent noiseless signals and diagrams. Deno-MAE achieves state-of-the-art accuracy in automatic modulation classification tasks with significantly fewer training samples, demonstrating a 10% reduction in unlabeled pretraining data and a 3% reduction in labeled fine-tuning data compared to existing approaches. Moreover, our model exhibits robust performance across varying signal-to-noise ratios (SNRs) and supports extrapolation on unseen lower SNRs. The results indicate that DenoMAE is an efficient, flexible, and data-efficient solution for denoising and classifying modulation signals in challenging noise-intensive environments.
2501.11540
A Hands-free Spatial Selection and Interaction Technique using Gaze and Blink Input with Blink Prediction for Extended Reality
cs.HC cs.LG
Gaze-based interaction techniques have created significant interest in the field of spatial interaction. Many of these methods require additional input modalities, such as hand gestures (e.g., gaze coupled with pinch). Those can be uncomfortable and difficult to perform in public or limited spaces, and pose challenges for users who are unable to execute pinch gestures. To address these aspects, we propose a novel, hands-free Gaze+Blink interaction technique that leverages the user's gaze and intentional eye blinks. This technique enables users to perform selections by executing intentional blinks. It facilitates continuous interactions, such as scrolling or drag-and-drop, through eye blinks coupled with head movements. So far, this concept has not been explored for hands-free spatial interaction techniques. We evaluated the performance and user experience (UX) of our Gaze+Blink method with two user studies and compared it with Gaze+Pinch in a realistic user interface setup featuring common menu interaction tasks. Study 1 demonstrated that while Gaze+Blink achieved comparable selection speeds, it was prone to accidental selections resulting from unintentional blinks. In Study 2 we explored an enhanced technique employing a deep learning algorithms for filtering out unintentional blinks.
2501.11542
DLinear-based Prediction of Remaining Useful Life of Lithium-Ion Batteries: Feature Engineering through Explainable Artificial Intelligence
eess.SY cs.LG cs.SY
Accurate prediction of the Remaining Useful Life (RUL) of lithium-ion batteries is essential for ensuring safety, reducing maintenance costs, and optimizing usage. However, predicting RUL is challenging due to the nonlinear characteristics of the degradation caused by complex chemical reactions. Machine learning allows precise predictions by learning the latent functions of degradation relationships based on cycling behavior. This study introduces an accurate RUL prediction approach based on feature engineering and DLinear, applied to the dataset from NASA's Prognostics Center of Excellence. Among the 20 features generated from current, voltage, temperature, and time provided in this dataset, key features contributing to degradation are selected using Pearson correlation coefficient and Shapley values. Shapley value-based feature selection effectively reflects cell-to-cell variability, showing similar importance rankings across all cells. The DLinear-based RUL prediction using key features efficiently captures the time-series trend, demonstrating significantly better performance compared to Long Short-Term Memory and Transformer models.
2501.11549
Whose Boat Does it Float? Improving Personalization in Preference Tuning via Inferred User Personas
cs.CL
LLMs are tuned to follow instructions (aligned) by learning which of two outputs users prefer for a prompt. However, this preference data format does not convey why users prefer responses that are chosen or rejected, so LLMs trained on these datasets cannot tailor responses to varied user needs. To surface these parameters of personalization, we apply abductive reasoning to preference data, inferring needs and interests of users, i.e. personas, that may prefer each output. We test this idea in two steps: Persona Inference (PI)-abductively inferring personas of users who prefer chosen or rejected outputs-and Persona Tailoring (PT)-training models to tailor responses to personas from PI. We find: 1) LLMs infer personas accurately explaining why different users may prefer both chosen or rejected outputs; 2) Training on preference data augmented with PI personas via PT boosts personalization, enabling models to support user-written personas; and 3) Rejected response personas form harder personalization evaluations, showing PT better aids users with uncommon preferences versus typical alignment methods. We argue for an abductive view of preferences for personalization, asking not only which response is better but when, why, and for whom.
2501.11551
PIKE-RAG: sPecIalized KnowledgE and Rationale Augmented Generation
cs.CL
Despite notable advancements in Retrieval-Augmented Generation (RAG) systems that expand large language model (LLM) capabilities through external retrieval, these systems often struggle to meet the complex and diverse needs of real-world industrial applications. The reliance on retrieval alone proves insufficient for extracting deep, domain-specific knowledge performing in logical reasoning from specialized corpora. To address this, we introduce sPecIalized KnowledgE and Rationale Augmentation Generation (PIKE-RAG), focusing on extracting, understanding, and applying specialized knowledge, while constructing coherent rationale to incrementally steer LLMs toward accurate responses. Recognizing the diverse challenges of industrial tasks, we introduce a new paradigm that classifies tasks based on their complexity in knowledge extraction and application, allowing for a systematic evaluation of RAG systems' problem-solving capabilities. This strategic approach offers a roadmap for the phased development and enhancement of RAG systems, tailored to meet the evolving demands of industrial applications. Furthermore, we propose knowledge atomizing and knowledge-aware task decomposition to effectively extract multifaceted knowledge from the data chunks and iteratively construct the rationale based on original query and the accumulated knowledge, respectively, showcasing exceptional performance across various benchmarks.
2501.11553
Clinically Ready Magnetic Microrobots for Targeted Therapies
cs.RO cond-mat.mtrl-sci cs.SY eess.SY physics.app-ph physics.bio-ph physics.med-ph
Systemic drug administration often causes off-target effects limiting the efficacy of advanced therapies. Targeted drug delivery approaches increase local drug concentrations at the diseased site while minimizing systemic drug exposure. We present a magnetically guided microrobotic drug delivery system capable of precise navigation under physiological conditions. This platform integrates a clinical electromagnetic navigation system, a custom-designed release catheter, and a dissolvable capsule for accurate therapeutic delivery. In vitro tests showed precise navigation in human vasculature models, and in vivo experiments confirmed tracking under fluoroscopy and successful navigation in large animal models. The microrobot balances magnetic material concentration, contrast agent loading, and therapeutic drug capacity, enabling effective hosting of therapeutics despite the integration complexity of its components, offering a promising solution for precise targeted drug delivery.
2501.11554
Event-based vision for egomotion estimation using precise event timing
cs.CV cs.AR cs.RO
Egomotion estimation is crucial for applications such as autonomous navigation and robotics, where accurate and real-time motion tracking is required. However, traditional methods relying on inertial sensors are highly sensitive to external conditions, and suffer from drifts leading to large inaccuracies over long distances. Vision-based methods, particularly those utilising event-based vision sensors, provide an efficient alternative by capturing data only when changes are perceived in the scene. This approach minimises power consumption while delivering high-speed, low-latency feedback. In this work, we propose a fully event-based pipeline for egomotion estimation that processes the event stream directly within the event-based domain. This method eliminates the need for frame-based intermediaries, allowing for low-latency and energy-efficient motion estimation. We construct a shallow spiking neural network using a synaptic gating mechanism to convert precise event timing into bursts of spikes. These spikes encode local optical flow velocities, and the network provides an event-based readout of egomotion. We evaluate the network's performance on a dedicated chip, demonstrating strong potential for low-latency, low-power motion estimation. Additionally, simulations of larger networks show that the system achieves state-of-the-art accuracy in egomotion estimation tasks with event-based cameras, making it a promising solution for real-time, power-constrained robotics applications.
2501.11555
Beyond R-barycenters: an effective averaging method on Stiefel and Grassmann manifolds
stat.ML cs.LG
In this paper, the issue of averaging data on a manifold is addressed. While the Fr\'echet mean resulting from Riemannian geometry appears ideal, it is unfortunately not always available and often computationally very expensive. To overcome this, R-barycenters have been proposed and successfully applied to Stiefel and Grassmann manifolds. However, R-barycenters still suffer severe limitations as they rely on iterative algorithms and complicated operators. We propose simpler, yet efficient, barycenters that we call RL-barycenters. We show that, in the setting relevant to most applications, our framework yields astonishingly simple barycenters: arithmetic means projected onto the manifold. We apply this approach to the Stiefel and Grassmann manifolds. On simulated data, our approach is competitive with respect to existing averaging methods, while computationally cheaper.
2501.11557
Secure Resource Allocation via Constrained Deep Reinforcement Learning
cs.LG
The proliferation of Internet of Things (IoT) devices and the advent of 6G technologies have introduced computationally intensive tasks that often surpass the processing capabilities of user devices. Efficient and secure resource allocation in serverless multi-cloud edge computing environments is essential for supporting these demands and advancing distributed computing. However, existing solutions frequently struggle with the complexity of multi-cloud infrastructures, robust security integration, and effective application of traditional deep reinforcement learning (DRL) techniques under system constraints. To address these challenges, we present SARMTO, a novel framework that integrates an action-constrained DRL model. SARMTO dynamically balances resource allocation, task offloading, security, and performance by utilizing a Markov decision process formulation, an adaptive security mechanism, and sophisticated optimization techniques. Extensive simulations across varying scenarios, including different task loads, data sizes, and MEC capacities, show that SARMTO consistently outperforms five baseline approaches, achieving up to a 40% reduction in system costs and a 41.5% improvement in energy efficiency over state-of-the-art methods. These enhancements highlight SARMTO's potential to revolutionize resource management in intricate distributed computing environments, opening the door to more efficient and secure IoT and edge computing applications.
2501.11560
Explainable Lane Change Prediction for Near-Crash Scenarios Using Knowledge Graph Embeddings and Retrieval Augmented Generation
cs.LG cs.AI cs.CL cs.IR
Lane-changing maneuvers, particularly those executed abruptly or in risky situations, are a significant cause of road traffic accidents. However, current research mainly focuses on predicting safe lane changes. Furthermore, existing accident datasets are often based on images only and lack comprehensive sensory data. In this work, we focus on predicting risky lane changes using the CRASH dataset (our own collected dataset specifically for risky lane changes), and safe lane changes (using the HighD dataset). Then, we leverage KG and Bayesian inference to predict these maneuvers using linguistic contextual information, enhancing the model's interpretability and transparency. The model achieved a 91.5% f1-score with anticipation time extending to four seconds for risky lane changes, and a 90.0% f1-score for predicting safe lane changes with the same anticipation time. We validate our model by integrating it into a vehicle within the CARLA simulator in scenarios that involve risky lane changes. The model managed to anticipate sudden lane changes, thus providing automated vehicles with further time to plan and execute appropriate safe reactions. Finally, to enhance the explainability of our model, we utilize RAG to provide clear and natural language explanations for the given prediction.
2501.11561
Teaching Large Language Models to Regress Accurate Image Quality Scores using Score Distribution
cs.CV
With the rapid advancement of Multi-modal Large Language Models (MLLMs), MLLM-based Image Quality Assessment (IQA) methods have shown promising performance in linguistic quality description. However, current methods still fall short in accurately scoring image quality. In this work, we aim to leverage MLLMs to regress accurate quality scores. A key challenge is that the quality score is inherently continuous, typically modeled as a Gaussian distribution, whereas MLLMs generate discrete token outputs. This mismatch necessitates score discretization. Previous approaches discretize the mean score into a one-hot label, resulting in information loss and failing to capture inter-image relationships. We propose a distribution-based approach that discretizes the score distribution into a soft label. This method preserves the characteristics of the score distribution, achieving high accuracy and maintaining inter-image relationships. Moreover, to address dataset variation, where different IQA datasets exhibit various distributions, we introduce a fidelity loss based on Thurstone's model. This loss captures intra-dataset relationships, facilitating co-training across multiple IQA datasets. With these designs, we develop the distribution-based Depicted image Quality Assessment model for Score regression (DeQA-Score). Experiments across multiple benchmarks show that DeQA-Score stably outperforms baselines in score regression. Also, DeQA-Score can predict the score distribution that closely aligns with human annotations. Codes and model weights have been released in https://depictqa.github.io/deqa-score/.
2501.11568
Graph Defense Diffusion Model
cs.LG
Graph Neural Networks (GNNs) demonstrate significant potential in various applications but remain highly vulnerable to adversarial attacks, which can greatly degrade their performance. Existing graph purification methods attempt to address this issue by filtering attacked graphs; however, they struggle to effectively defend against multiple types of adversarial attacks simultaneously due to their limited flexibility, and they lack comprehensive modeling of graph data due to their heavy reliance on heuristic prior knowledge. To overcome these challenges, we propose a more versatile approach for defending against adversarial attacks on graphs. In this work, we introduce the Graph Defense Diffusion Model (GDDM), a flexible purification method that leverages the denoising and modeling capabilities of diffusion models. The iterative nature of diffusion models aligns well with the stepwise process of adversarial attacks, making them particularly suitable for defense. By iteratively adding and removing noise, GDDM effectively purifies attacked graphs, restoring their original structure and features. Our GDDM consists of two key components: (1) Graph Structure-Driven Refiner, which preserves the basic fidelity of the graph during the denoising process, and ensures that the generated graph remains consistent with the original scope; and (2) Node Feature-Constrained Regularizer, which removes residual impurities from the denoised graph, further enhances the purification effect. Additionally, we design tailored denoising strategies to handle different types of adversarial attacks, improving the model's adaptability to various attack scenarios. Extensive experiments conducted on three real-world datasets demonstrate that GDDM outperforms state-of-the-art methods in defending against a wide range of adversarial attacks, showcasing its robustness and effectiveness.
2501.11570
Uncertainty Estimation in the Real World: A Study on Music Emotion Recognition
cs.SD cs.IR cs.LG eess.AS
Any data annotation for subjective tasks shows potential variations between individuals. This is particularly true for annotations of emotional responses to musical stimuli. While older approaches to music emotion recognition systems frequently addressed this uncertainty problem through probabilistic modeling, modern systems based on neural networks tend to ignore the variability and focus only on predicting central tendencies of human subjective responses. In this work, we explore several methods for estimating not only the central tendencies of the subjective responses to a musical stimulus, but also for estimating the uncertainty associated with these responses. In particular, we investigate probabilistic loss functions and inference-time random sampling. Experimental results indicate that while the modeling of the central tendencies is achievable, modeling of the uncertainty in subjective responses proves significantly more challenging with currently available approaches even when empirical estimates of variations in the responses are available.
2501.11577
Rethinking Membership Inference Attacks Against Transfer Learning
cs.CR cs.LG
Transfer learning, successful in knowledge translation across related tasks, faces a substantial privacy threat from membership inference attacks (MIAs). These attacks, despite posing significant risk to ML model's training data, remain limited-explored in transfer learning. The interaction between teacher and student models in transfer learning has not been thoroughly explored in MIAs, potentially resulting in an under-examined aspect of privacy vulnerabilities within transfer learning. In this paper, we propose a new MIA vector against transfer learning, to determine whether a specific data point was used to train the teacher model while only accessing the student model in a white-box setting. Our method delves into the intricate relationship between teacher and student models, analyzing the discrepancies in hidden layer representations between the student model and its shadow counterpart. These identified differences are then adeptly utilized to refine the shadow model's training process and to inform membership inference decisions effectively. Our method, evaluated across four datasets in diverse transfer learning tasks, reveals that even when an attacker only has access to the student model, the teacher model's training data remains susceptible to MIAs. We believe our work unveils the unexplored risk of membership inference in transfer learning.
2501.11584
GCSAM: Gradient Centralized Sharpness Aware Minimization
cs.LG
The generalization performance of deep neural networks (DNNs) is a critical factor in achieving robust model behavior on unseen data. Recent studies have highlighted the importance of sharpness-based measures in promoting generalization by encouraging convergence to flatter minima. Among these approaches, Sharpness-Aware Minimization (SAM) has emerged as an effective optimization technique for reducing the sharpness of the loss landscape, thereby improving generalization. However, SAM's computational overhead and sensitivity to noisy gradients limit its scalability and efficiency. To address these challenges, we propose Gradient-Centralized Sharpness-Aware Minimization (GCSAM), which incorporates Gradient Centralization (GC) to stabilize gradients and accelerate convergence. GCSAM normalizes gradients before the ascent step, reducing noise and variance, and improving stability during training. Our evaluations indicate that GCSAM consistently outperforms SAM and the Adam optimizer in terms of generalization and computational efficiency. These findings demonstrate GCSAM's effectiveness across diverse domains, including general and medical imaging tasks.
2501.11586
Compressibility Analysis for the differentiable shift-variant Filtered Backprojection Model
cs.CV eess.IV
The differentiable shift-variant filtered backprojection (FBP) model enables the reconstruction of cone-beam computed tomography (CBCT) data for any non-circular trajectories. This method employs deep learning technique to estimate the redundancy weights required for reconstruction, given knowledge of the specific trajectory at optimization time. However, computing the redundancy weight for each projection remains computationally intensive. This paper presents a novel approach to compress and optimize the differentiable shift-variant FBP model based on Principal Component Analysis (PCA). We apply PCA to the redundancy weights learned from sinusoidal trajectory projection data, revealing significant parameter redundancy in the original model. By integrating PCA directly into the differentiable shift-variant FBP reconstruction pipeline, we develop a method that decomposes the redundancy weight layer parameters into a trainable eigenvector matrix, compressed weights, and a mean vector. This innovative technique achieves a remarkable 97.25% reduction in trainable parameters without compromising reconstruction accuracy. As a result, our algorithm significantly decreases the complexity of the differentiable shift-variant FBP model and greatly improves training speed. These improvements make the model substantially more practical for real-world applications.
2501.11587
Recurrent Diffusion for Large-Scale Parameter Generation
cs.LG cs.AI
Parameter generation has long struggled to match the scale of today large vision and language models, curbing its broader utility. In this paper, we introduce Recurrent Diffusion for Large Scale Parameter Generation (RPG), a novel framework that generates full neural network parameters up to hundreds of millions on a single GPU. Our approach first partitions a networks parameters into non-overlapping tokens, each corresponding to a distinct portion of the model. A recurrent mechanism then learns the inter token relationships, producing prototypes which serve as conditions for a diffusion process that ultimately synthesizes the full parameters. Across a spectrum of architectures and tasks including ResNets, ConvNeXts and ViTs on ImageNet 1K and COCO, and even LoRA based LLMs RPG achieves performance on par with fully trained networks while avoiding excessive memory overhead. Notably, it generalizes beyond its training set to generate valid parameters for previously unseen tasks, highlighting its flexibility in dynamic and open ended scenarios. By overcoming the longstanding memory and scalability barriers, RPG serves as a critical advance in AI generating AI, potentially enabling efficient weight generation at scales previously deemed infeasible.
2501.11592
Training-free Ultra Small Model for Universal Sparse Reconstruction in Compressed Sensing
cs.LG cs.AI cs.CL
Pre-trained large models attract widespread attention in recent years, but they face challenges in applications that require high interpretability or have limited resources, such as physical sensing, medical imaging, and bioinformatics. Compressed Sensing (CS) is a well-proved theory that drives many recent breakthroughs in these applications. However, as a typical under-determined linear system, CS suffers from excessively long sparse reconstruction times when using traditional iterative methods, particularly with large-scale data. Current AI methods like deep unfolding fail to substitute them because pre-trained models exhibit poor generality beyond their training conditions and dataset distributions, or lack interpretability. Instead of following the big model fervor, this paper proposes ultra-small artificial neural models called coefficients learning (CL), enabling training-free and rapid sparse reconstruction while perfectly inheriting the generality and interpretability of traditional iterative methods, bringing new feature of incorporating prior knowledges. In CL, a signal of length $n$ only needs a minimal of $n$ trainable parameters. A case study model called CLOMP is implemented for evaluation. Experiments are conducted on both synthetic and real one-dimensional and two-dimensional signals, demonstrating significant improvements in efficiency and accuracy. Compared to representative iterative methods, CLOMP improves efficiency by 100 to 1000 folds for large-scale data. Test results on eight diverse image datasets indicate that CLOMP improves structural similarity index by 292%, 98%, 45% for sampling rates of 0.1, 0.3, 0.5, respectively. We believe this method can truly usher CS reconstruction into the AI era, benefiting countless under-determined linear systems that rely on sparse solution.
2501.11593
Optimal User and Target Scheduling, User-Target Pairing, and Low-Resolution Phase-Only Beamforming for ISAC Systems
eess.SP cs.IT cs.NI math.IT
We investigate the joint user and target scheduling, user-target pairing, and low-resolution phase-only beamforming design for integrated sensing and communications (ISAC). Scheduling determines which users and targets are served, while pairing specifies which users and targets are grouped into pairs. Additionally, the beamformers are designed using few-bit constant-modulus phase shifts. This resource allocation problem is a nonconvex mixed-integer nonlinear program (MINLP) and challenging to solve. To address it, we propose an exact mixed-integer linear program (MILP) reformulation, which leads to a globally optimal solution. Our results demonstrate the superiority of an optimal joint design compared to heuristic stage-wise approaches, which are highly sensitive to scenario characteristics.
2501.11597
Fairness Testing through Extreme Value Theory
cs.SE cs.AI cs.CY cs.LG
Data-driven software is increasingly being used as a critical component of automated decision-support systems. Since this class of software learns its logic from historical data, it can encode or amplify discriminatory practices. Previous research on algorithmic fairness has focused on improving average-case fairness. On the other hand, fairness at the extreme ends of the spectrum, which often signifies lasting and impactful shifts in societal attitudes, has received significantly less emphasis. Leveraging the statistics of extreme value theory (EVT), we propose a novel fairness criterion called extreme counterfactual discrimination (ECD). This criterion estimates the worst-case amounts of disadvantage in outcomes for individuals solely based on their memberships in a protected group. Utilizing tools from search-based software engineering and generative AI, we present a randomized algorithm that samples a statistically significant set of points from the tail of ML outcome distributions even if the input dataset lacks a sufficient number of relevant samples. We conducted several experiments on four ML models (deep neural networks, logistic regression, and random forests) over 10 socially relevant tasks from the literature on algorithmic fairness. First, we evaluate the generative AI methods and find that they generate sufficient samples to infer valid EVT distribution in 95% of cases. Remarkably, we found that the prevalent bias mitigators reduce the average-case discrimination but increase the worst-case discrimination significantly in 5% of cases. We also observed that even the tail-aware mitigation algorithm -- MiniMax-Fairness -- increased the worst-case discrimination in 30% of cases. We propose a novel ECD-based mitigator that improves fairness in the tail in 90% of cases with no degradation of the average-case discrimination.
2501.11599
SR-FoT: A Syllogistic-Reasoning Framework of Thought for Large Language Models Tackling Knowledge-based Reasoning Tasks
cs.AI cs.CL
Deductive reasoning is a crucial logical capability that assists us in solving complex problems based on existing knowledge. Although augmented by Chain-of-Thought prompts, Large Language Models (LLMs) might not follow the correct reasoning paths. Enhancing the deductive reasoning abilities of LLMs, and leveraging their extensive built-in knowledge for various reasoning tasks, remains an open question. Attempting to mimic the human deductive reasoning paradigm, we propose a multi-stage Syllogistic-Reasoning Framework of Thought (SR-FoT) that enables LLMs to perform syllogistic deductive reasoning to handle complex knowledge-based reasoning tasks. Our SR-FoT begins by interpreting the question and then uses the interpretation and the original question to propose a suitable major premise. It proceeds by generating and answering minor premise questions in two stages to match the minor premises. Finally, it guides LLMs to use the previously generated major and minor premises to perform syllogistic deductive reasoning to derive the answer to the original question. Extensive and thorough experiments on knowledge-based reasoning tasks have demonstrated the effectiveness and advantages of our SR-FoT.
2501.11605
Bootstrapping Social Networks: Lessons from Bluesky Starter Packs
cs.SI cs.NI
Microblogging is a crucial mode of online communication. However, launching a new microblogging platform remains challenging, largely due to network effects. This has resulted in entrenched (and undesirable) dominance by established players, such as X/Twitter. To overcome these network effects, Bluesky, an emerging microblogging platform, introduced starter packs -- curated lists of accounts that users can follow with a single click. We ask if starter packs have the potential to tackle the critical problem of social bootstrapping in new online social networks? This paper is the first to address this question: we asses whether starter packs have been indeed helpful in supporting Bluesky growth. Our dataset includes $25.05 \times 10^6$ users and $335.42 \times 10^3$ starter packs with $1.73 \times 10^6$ members, covering the entire lifecycle of Bluesky. We study the usage of these starter packs, their ability to drive network and activity growth, and their potential downsides. We also quantify the benefits of starter packs for members and creators on user visibility and activity while identifying potential challenges. By evaluating starter packs' effectiveness and limitations, we contribute to the broader discourse on platform growth strategies and competitive innovation in the social media landscape.
2501.11613
Conversation Routines: A Prompt Engineering Framework for Task-Oriented Dialog Systems
cs.CL cs.AI cs.ET cs.HC cs.PL
This study introduces Conversation Routines (CR), a structured prompt engineering framework for developing task-oriented dialog systems using Large Language Models (LLMs). While LLMs demonstrate remarkable natural language understanding capabilities, engineering them to reliably execute complex business workflows remains challenging. The proposed CR framework enables the development of Conversation Agentic Systems (CAS) through natural language specifications, embedding task-oriented logic within LLM prompts. This approach provides a systematic methodology for designing and implementing complex conversational workflows while maintaining behavioral consistency. We demonstrate the framework's effectiveness through two proof-of-concept implementations: a Train Ticket Booking System and an Interactive Troubleshooting Copilot. These case studies validate CR's capability to encode sophisticated behavioral patterns and decision logic while preserving natural conversational flexibility. Results show that CR enables domain experts to design conversational workflows in natural language while leveraging custom functions (tools) developed by software engineers, creating an efficient division of responsibilities where developers focus on core API implementation and domain experts handle conversation design. While the framework shows promise in accessibility and adaptability, we identify key challenges including computational overhead, non-deterministic behavior, and domain-specific logic optimization. Future research directions include CR evaluation methods based on prompt engineering frameworks driven by goal-oriented grading criteria, improving scalability for complex multi-agent interactions, and enhancing system robustness to address the identified limitations across diverse business applications.
2501.11621
Trojan Detection Through Pattern Recognition for Large Language Models
cs.CL cs.LG
Trojan backdoors can be injected into large language models at various stages, including pretraining, fine-tuning, and in-context learning, posing a significant threat to the model's alignment. Due to the nature of causal language modeling, detecting these triggers is challenging given the vast search space. In this study, we propose a multistage framework for detecting Trojan triggers in large language models consisting of token filtration, trigger identification, and trigger verification. We discuss existing trigger identification methods and propose two variants of a black-box trigger inversion method that rely on output logits, utilizing beam search and greedy decoding respectively. We show that the verification stage is critical in the process and propose semantic-preserving prompts and special perturbations to differentiate between actual Trojan triggers and other adversarial strings that display similar characteristics. The evaluation of our approach on the TrojAI and RLHF poisoned model datasets demonstrates promising results.
2501.11622
Causal Learning for Heterogeneous Subgroups Based on Nonlinear Causal Kernel Clustering
cs.LG stat.ML
Due to the challenge posed by multi-source and heterogeneous data collected from diverse environments, causal relationships among features can exhibit variations influenced by different time spans, regions, or strategies. This diversity makes a single causal model inadequate for accurately representing complex causal relationships in all observational data, a crucial consideration in causal learning. To address this challenge, the nonlinear Causal Kernel Clustering method is introduced for heterogeneous subgroup causal learning, highlighting variations in causal relationships across diverse subgroups. The main component for clustering heterogeneous subgroups lies in the construction of the $u$-centered sample mapping function with the property of unbiased estimation, which assesses the differences in potential nonlinear causal relationships in various samples and supported by causal identifiability theory. Experimental results indicate that the method performs well in identifying heterogeneous subgroups and enhancing causal learning, leading to a reduction in prediction error.
2501.11623
Early evidence of how LLMs outperform traditional systems on OCR/HTR tasks for historical records
cs.CV cs.AI cs.LG
We explore the ability of two LLMs -- GPT-4o and Claude Sonnet 3.5 -- to transcribe historical handwritten documents in a tabular format and compare their performance to traditional OCR/HTR systems: EasyOCR, Keras, Pytesseract, and TrOCR. Considering the tabular form of the data, two types of experiments are executed: one where the images are split line by line and the other where the entire scan is used as input. Based on CER and BLEU, we demonstrate that LLMs outperform the conventional OCR/HTR methods. Moreover, we also compare the evaluated CER and BLEU scores to human evaluations to better judge the outputs of whole-scan experiments and understand influential factors for CER and BLEU. Combining judgments from all the evaluation metrics, we conclude that two-shot GPT-4o for line-by-line images and two-shot Claude Sonnet 3.5 for whole-scan images yield the transcriptions of the historical records most similar to the ground truth.
2501.11626
DRL-Based Maximization of the Sum Cross-Layer Achievable Rate for Networks Under Jamming
eess.SY cs.SY
In quasi-static wireless networks characterized by infrequent changes in the transmission schedules of user equipment (UE), malicious jammers can easily deteriorate network performance. Accordingly, a key challenge in these networks is managing channel access amidst jammers and under dynamic channel conditions. In this context, we propose a robust learning-based mechanism for channel access in multi-cell quasi-static networks under jamming. The network comprises multiple legitimate UEs, including predefined UEs (pUEs) with stochastic predefined schedules and an intelligent UE (iUE) with an undefined transmission schedule, all transmitting over a shared, time-varying uplink channel. Jammers transmit unwanted packets to disturb the pUEs' and the iUE's communication. The iUE's learning process is based on the deep reinforcement learning (DRL) framework, utilizing a residual network (ResNet)-based deep Q-Network (DQN). To coexist in the network and maximize the network's sum cross-layer achievable rate (SCLAR), the iUE must learn the unknown network dynamics while concurrently adapting to dynamic channel conditions. Our simulation results reveal that, with properly defined state space, action space, and rewards in DRL, the iUE can effectively coexist in the network, maximizing channel utilization and the network's SCLAR by judiciously selecting transmission time slots and thus avoiding collisions and jamming.
2501.11628
Investigating the Scalability of Approximate Sparse Retrieval Algorithms to Massive Datasets
cs.IR
Learned sparse text embeddings have gained popularity due to their effectiveness in top-k retrieval and inherent interpretability. Their distributional idiosyncrasies, however, have long hindered their use in real-world retrieval systems. That changed with the recent development of approximate algorithms that leverage the distributional properties of sparse embeddings to speed up retrieval. Nonetheless, in much of the existing literature, evaluation has been limited to datasets with only a few million documents such as MSMARCO. It remains unclear how these systems behave on much larger datasets and what challenges lurk in larger scales. To bridge that gap, we investigate the behavior of state-of-the-art retrieval algorithms on massive datasets. We compare and contrast the recently-proposed Seismic and graph-based solutions adapted from dense retrieval. We extensively evaluate Splade embeddings of 138M passages from MsMarco-v2 and report indexing time and other efficiency and effectiveness metrics.
2501.11631
Noise-Agnostic Multitask Whisper Training for Reducing False Alarm Errors in Call-for-Help Detection
cs.SD cs.AI eess.AS
Keyword spotting is often implemented by keyword classifier to the encoder in acoustic models, enabling the classification of predefined or open vocabulary keywords. Although keyword spotting is a crucial task in various applications and can be extended to call-for-help detection in emergencies, however, the previous method often suffers from scalability limitations due to retraining required to introduce new keywords or adapt to changing contexts. We explore a simple yet effective approach that leverages off-the-shelf pretrained ASR models to address these challenges, especially in call-for-help detection scenarios. Furthermore, we observed a substantial increase in false alarms when deploying call-for-help detection system in real-world scenarios due to noise introduced by microphones or different environments. To address this, we propose a novel noise-agnostic multitask learning approach that integrates a noise classification head into the ASR encoder. Our method enhances the model's robustness to noisy environments, leading to a significant reduction in false alarms and improved overall call-for-help performance. Despite the added complexity of multitask learning, our approach is computationally efficient and provides a promising solution for call-for-help detection in real-world scenarios.
2501.11632
Biomedical Knowledge Graph: A Survey of Domains, Tasks, and Real-World Applications
cs.CL cs.AI cs.CE cs.IR
Biomedical knowledge graphs (BKGs) have emerged as powerful tools for organizing and leveraging the vast and complex data found across the biomedical field. Yet, current reviews of BKGs often limit their scope to specific domains or methods, overlooking the broader landscape and the rapid technological progress reshaping it. In this survey, we address this gap by offering a systematic review of BKGs from three core perspectives: domains, tasks, and applications. We begin by examining how BKGs are constructed from diverse data sources, including molecular interactions, pharmacological datasets, and clinical records. Next, we discuss the essential tasks enabled by BKGs, focusing on knowledge management, retrieval, reasoning, and interpretation. Finally, we highlight real-world applications in precision medicine, drug discovery, and scientific research, illustrating the translational impact of BKGs across multiple sectors. By synthesizing these perspectives into a unified framework, this survey not only clarifies the current state of BKG research but also establishes a foundation for future exploration, enabling both innovative methodological advances and practical implementations.
2501.11633
PSO-based Sliding Mode Current Control of Grid-Forming Inverter in Rotating Frame
eess.SY cs.SY
The Grid-Forming Inverter (GFMI) is an emerging topic that is attracting significant attention from both academic and industrial communities, particularly in the area of control design. The Decoupled Average Model-based Sliding Mode Current Controller (DAM-SMC) has been used to address the need such as fast response, fixed switching frequency, and no overshoot to avoid exceeding current limits. Typically, the control parameters for DAM-SMC are chosen based on expert knowledge and certain assumptions. However, these parameters may not achieve optimized performance due to system dynamics and uncertainties. To address this, this paper proposes a Particle Swarm Optimization (PSO)-based DAM-SMC controller, which inherits the control laws from DAM-SMC but optimizes the control parameters offline using PSO. The main goal is to reduce chattering and achieve smaller tracking errors. The proposed method is compared with other metaheuristic optimization algorithms, such as Genetic Algorithm (GA) and Simulated Annealing (SA). Simulations are performed in MATLAB/Simulink across various scenarios to evaluate the effectiveness of the proposed controller. The proposed approach achieves a substantial reduction in convergence time, decreasing it by 86.36% compared to the GA and by 88.89% compared to SA. Furthermore, the tracking error is reduced by 11.61% compared to the conventional DAM-SMC algorithm. The robustness of the proposed method is validated under critical conditions, where plant and control model parameters varied by up to 40%.
2501.11636
Characterization of the Arithmetic Complexity of the Secrecy Capacity of Fast-Fading Gaussian Channels
cs.IT math.IT
This paper studies the computability of the secrecy capacity of fast-fading wiretap channels from an algorithmic perspective, examining whether it can be computed algorithmically or not. To address this question, the concept of Turing machines is used, which establishes fundamental performance limits of digital computers. It is shown that certain computable continuous fading probability distribution functions yield secrecy capacities that are non-computable numbers. Additionally, we assess the secrecy capacity's classification within the arithmetical hierarchy, revealing the absence of computable achievability and converse bounds.
2501.11638
Class Imbalance in Anomaly Detection: Learning from an Exactly Solvable Model
cs.LG cond-mat.dis-nn stat.ML
Class imbalance (CI) is a longstanding problem in machine learning, slowing down training and reducing performances. Although empirical remedies exist, it is often unclear which ones work best and when, due to the lack of an overarching theory. We address a common case of imbalance, that of anomaly (or outlier) detection. We provide a theoretical framework to analyze, interpret and address CI. It is based on an exact solution of the teacher-student perceptron model, through replica theory. Within this framework, one can distinguish several sources of CI: either intrinsic, train or test imbalance. Our analysis reveals that the optimal train imbalance is generally different from 50%, with a non trivial dependence on the intrinsic imbalance, the abundance of data and on the noise in the learning. Moreover, there is a crossover between a small noise training regime where results are independent of the noise level to a high noise regime where performances quickly degrade with noise. Our results challenge some of the conventional wisdom on CI and offer practical guidelines to address it.
2501.11639
StAyaL | Multilingual Style Transfer
cs.CL cs.AI
Stylistic text generation plays a vital role in enhancing communication by reflecting the nuances of individual expression. This paper presents a novel approach for generating text in a specific speaker's style across different languages. We show that by leveraging only 100 lines of text, an individuals unique style can be captured as a high-dimensional embedding, which can be used for both text generation and stylistic translation. This methodology breaks down the language barrier by transferring the style of a speaker between languages. The paper is structured into three main phases: augmenting the speaker's data with stylistically consistent external sources, separating style from content using machine learning and deep learning techniques, and generating an abstract style profile by mean pooling the learned embeddings. The proposed approach is shown to be topic-agnostic, with test accuracy and F1 scores of 74.9% and 0.75, respectively. The results demonstrate the potential of the style profile for multilingual communication, paving the way for further applications in personalized content generation and cross-linguistic stylistic transfer.
2501.11641
A Common Ancestor of PDL, Conjunctive Queries, and Unary Negation First-order
cs.LO cs.DB
We introduce and study UCPDL+, a family of expressive logics rooted in Propositional Dynamic Logic (PDL) with converse (CPDL) and universal modality (UCPDL). In terms of expressive power, UCPDL+ strictly contains PDL extended with intersection and converse (a.k.a. ICPDL), as well as Conjunctive Queries (CQ), Conjunctive Regular Path Queries (CRPQ), or some known extensions thereof (Regular Queries and CQPDL). Further, it is equivalent to the extension of the unary-negation fragment of first-order logic (UNFO) with unary transitive closure, which we denote by UNFO*, which in turn strictly contains a previously studied extension of UNFO with regular expressions known as UNFO^reg. We investigate the expressive power, indistinguishability via bisimulations, satisfiability, and model checking for UCPDL+ and CPDL+. We argue that natural subclasses of CPDL+ can be defined in terms of the tree-width of the underlying graphs of the formulas. We show that the class of CPDL+ formulas of tree-width 2 is equivalent to ICPDL, and that it also coincides with CPDL+ formulas of tree-width 1. However, beyond tree-width 2, incrementing the tree-width strictly increases the expressive power. We characterize the expressive power for every class of fixed tree-width formulas in terms of a bisimulation game with pebbles. Based on this characterization, we show that CPDL+ has a tree-like model property. We prove that the satisfiability problem for UCPDL+ is decidable in 2ExpTime, coinciding with the complexity of ICPDL. As a consequence, the satisfiability problem for UNFO* is shown to be 2ExpTime-complete as well. We also exhibit classes for which satisfiability is reduced to ExpTime. Finally, we establish that the model checking problem for fixed tree-width formulas is in PTime, contrary to the full class CPDL+.
2501.11651
Advancing Language Model Reasoning through Reinforcement Learning and Inference Scaling
cs.LG cs.CL
Large language models (LLMs) have demonstrated remarkable capabilities in complex reasoning tasks. However, existing approaches mainly rely on imitation learning and struggle to achieve effective test-time scaling. While reinforcement learning (RL) holds promise for enabling self-exploration and learning from feedback, recent attempts yield only modest improvements in complex reasoning. In this paper, we present T1 to scale RL by encouraging exploration and understand inference scaling. We first initialize the LLM using synthesized chain-of-thought data that integrates trial-and-error and self-verification. To scale RL training, we promote increased sampling diversity through oversampling. We further employ an entropy bonus as an auxiliary loss, alongside a dynamic anchor for regularization to facilitate reward optimization. We demonstrate that T1 with open LLMs as its base exhibits inference scaling behavior and achieves superior performance on challenging math reasoning benchmarks. For example, T1 with Qwen2.5-32B as the base model outperforms the recent Qwen QwQ-32B-Preview model on MATH500, AIME2024, and Omni-math-500. More importantly, we present a simple strategy to examine inference scaling, where increased inference budgets directly lead to T1's better performance without any additional verification. We will open-source the T1 models and the data used to train them at \url{https://github.com/THUDM/T1}.
2501.11653
Dynamic Scene Understanding from Vision-Language Representations
cs.CV cs.LG
Images depicting complex, dynamic scenes are challenging to parse automatically, requiring both high-level comprehension of the overall situation and fine-grained identification of participating entities and their interactions. Current approaches use distinct methods tailored to sub-tasks such as Situation Recognition and detection of Human-Human and Human-Object Interactions. However, recent advances in image understanding have often leveraged web-scale vision-language (V&L) representations to obviate task-specific engineering. In this work, we propose a framework for dynamic scene understanding tasks by leveraging knowledge from modern, frozen V&L representations. By framing these tasks in a generic manner - as predicting and parsing structured text, or by directly concatenating representations to the input of existing models - we achieve state-of-the-art results while using a minimal number of trainable parameters relative to existing approaches. Moreover, our analysis of dynamic knowledge of these representations shows that recent, more powerful representations effectively encode dynamic scene semantics, making this approach newly possible.
2501.11655
KKL Observer Synthesis for Nonlinear Systems via Physics-Informed Learning
eess.SY cs.LG cs.SY
This paper proposes a novel learning approach for designing Kazantzis-Kravaris/Luenberger (KKL) observers for autonomous nonlinear systems. The design of a KKL observer involves finding an injective map that transforms the system state into a higher-dimensional observer state, whose dynamics is linear and stable. The observer's state is then mapped back to the original system coordinates via the inverse map to obtain the state estimate. However, finding this transformation and its inverse is quite challenging. We propose to sequentially approximate these maps by neural networks that are trained using physics-informed learning. We generate synthetic data for training by numerically solving the system and observer dynamics. Theoretical guarantees for the robustness of state estimation against approximation error and system uncertainties are provided. Additionally, a systematic method for optimizing observer performance through parameter selection is presented. The effectiveness of the proposed approach is demonstrated through numerical simulations on benchmark examples and its application to sensor fault detection and isolation in a network of Kuramoto oscillators using learned KKL observers.
2501.11657
Classification of HI Galaxy Profiles Using Unsupervised Learning and Convolutional Neural Networks: A Comparative Analysis and Methodological Cases of Studies
astro-ph.GA cs.LG
Hydrogen, the most abundant element in the universe, is crucial for understanding galaxy formation and evolution. The 21 cm neutral atomic hydrogen - HI spectral line maps the gas kinematics within galaxies, providing key insights into interactions, galactic structure, and star formation processes. With new radio instruments, the volume and complexity of data is increasing. To analyze and classify integrated HI spectral profiles in a efficient way, this work presents a framework that integrates Machine Learning techniques, combining unsupervised methods and CNNs. To this end, we apply our framework to a selected subsample of 318 spectral HI profiles of the CIG and 30.780 profiles from the Arecibo Legacy Fast ALFA Survey catalogue. Data pre-processing involved the Busyfit package and iterative fitting with polynomial, Gaussian, and double-Lorentzian models. Clustering methods, including K-means, spectral clustering, DBSCAN, and agglomerative clustering, were used for feature extraction and to bootstrap classification we applied K-NN, SVM, and Random Forest classifiers, optimizing accuracy with CNN. Additionally, we introduced a 2D model of the profiles to enhance classification by adding dimensionality to the data. Three 2D models were generated based on transformations and normalised versions to quantify the level of asymmetry. These methods were tested in a previous analytical classification study conducted by the Analysis of the Interstellar Medium in Isolated Galaxies group. This approach enhances classification accuracy and aims to establish a methodology that could be applied to data analysis in future surveys conducted with the Square Kilometre Array (SKA), currently under construction. All materials, code, and models have been made publicly available in an open-access repository, adhering to FAIR principles.
2501.11671
Exploring Preference-Guided Diffusion Model for Cross-Domain Recommendation
cs.IR
Cross-domain recommendation (CDR) has been proven as a promising way to alleviate the cold-start issue, in which the most critical problem is how to draw an informative user representation in the target domain via the transfer of user preference existing in the source domain. Prior efforts mostly follow the embedding-and-mapping paradigm, which first integrate the preference into user representation in the source domain, and then perform a mapping function on this representation to the target domain. However, they focus on mapping features across domains, neglecting to explicitly model the preference integration process, which may lead to learning coarse user representation. Diffusion models (DMs), which contribute to more accurate user/item representations due to their explicit information injection capability, have achieved promising performance in recommendation systems. Nevertheless, these DMs-based methods cannot directly account for valuable user preference in other domains, leading to challenges in adapting to the transfer of preference for cold-start users. Consequently, the feasibility of DMs for CDR remains underexplored. To this end, we explore to utilize the explicit information injection capability of DMs for user preference integration and propose a Preference-Guided Diffusion Model for CDR to cold-start users, termed as DMCDR. Specifically, we leverage a preference encoder to establish the preference guidance signal with the user's interaction history in the source domain. Then, we explicitly inject the preference guidance signal into the user representation step by step to guide the reverse process, and ultimately generate the personalized user representation in the target domain, thus achieving the transfer of user preference across domains. Furthermore, we comprehensively explore the impact of six DMs-based variants on CDR.