id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
2502.14619
Reward Models Identify Consistency, Not Causality
cs.LG cs.AI cs.CL
Reward models (RMs) play a crucial role in aligning large language models (LLMs) with human preferences and enhancing reasoning quality. Traditionally, RMs are trained to rank candidate outputs based on their correctness and coherence. However, in this work, we present several surprising findings that challenge common assumptions about RM behavior. Our analysis reveals that state-of-the-art reward models prioritize structural consistency over causal correctness. Specifically, removing the problem statement has minimal impact on reward scores, whereas altering numerical values or disrupting the reasoning flow significantly affects RM outputs. Furthermore, RMs exhibit a strong dependence on complete reasoning trajectories truncated or incomplete steps lead to significant variations in reward assignments, indicating that RMs primarily rely on learned reasoning patterns rather than explicit problem comprehension. These findings hold across multiple architectures, datasets, and tasks, leading to three key insights: (1) RMs primarily assess coherence rather than true reasoning quality; (2) The role of explicit problem comprehension in reward assignment is overstated; (3) Current RMs may be more effective at ranking responses than verifying logical validity. Our results suggest a fundamental limitation in existing reward modeling approaches, emphasizing the need for a shift toward causality-aware reward models that go beyond consistency-driven evaluation.
2502.14620
Exploring RWKV for Sentence Embeddings: Layer-wise Analysis and Baseline Comparison for Semantic Similarity
cs.CL cs.AI
This paper investigates the efficacy of RWKV, a novel language model architecture known for its linear attention mechanism, for generating sentence embeddings in a zero-shot setting. I conduct a layer-wise analysis to evaluate the semantic similarity captured by embeddings from different hidden layers of a pre-trained RWKV model. The performance is assessed on the Microsoft Research Paraphrase Corpus (MRPC) dataset using Spearman correlation and compared against a GloVe-based baseline. My results indicate that while RWKV embeddings capture some semantic relatedness, they underperform compared to the GloVe baseline in terms of Spearman correlation. I also analyze the inference time and GPU memory usage, highlighting the computational trade-offs associated with RWKV embeddings. The findings suggest that while RWKV offers potential advantages in terms of linear scaling, its zero-shot sentence embedding quality for semantic similarity tasks requires further investigation and potential task-specific fine-tuning to match or exceed simpler baselines.
2502.14625
Multi-Record Web Page Information Extraction From News Websites
cs.CL cs.IR
In this paper, we focused on the problem of extracting information from web pages containing many records, a task of growing importance in the era of massive web data. Recently, the development of neural network methods has improved the quality of information extraction from web pages. Nevertheless, most of the research and datasets are aimed at studying detailed pages. This has left multi-record "list pages" relatively understudied, despite their widespread presence and practical significance. To address this gap, we created a large-scale, open-access dataset specifically designed for list pages. This is the first dataset for this task in the Russian language. Our dataset contains 13,120 web pages with news lists, significantly exceeding existing datasets in both scale and complexity. Our dataset contains attributes of various types, including optional and multi-valued, providing a realistic representation of real-world list pages. These features make our dataset a valuable resource for studying information extraction from pages containing many records. Furthermore, we proposed our own multi-stage information extraction methods. In this work, we explore and demonstrate several strategies for applying MarkupLM to the specific challenges of multi-record web pages. Our experiments validate the advantages of our methods. By releasing our dataset to the public, we aim to advance the field of information extraction from multi-record pages.
2502.14627
ATRI: Mitigating Multilingual Audio Text Retrieval Inconsistencies by Reducing Data Distribution Errors
cs.SD cs.AI eess.AS
Multilingual audio-text retrieval (ML-ATR) is a challenging task that aims to retrieve audio clips or multilingual texts from databases. However, existing ML-ATR schemes suffer from inconsistencies for instance similarity matching across languages. We theoretically analyze the inconsistency in terms of both multilingual modal alignment direction error and weight error, and propose the theoretical weight error upper bound for quantifying the inconsistency. Based on the analysis of the weight error upper bound, we find that the inconsistency problem stems from the data distribution error caused by random sampling of languages. We propose a consistent ML-ATR scheme using 1-to-k contrastive learning and audio-English co-anchor contrastive learning, aiming to mitigate the negative impact of data distribution error on recall and consistency in ML-ATR. Experimental results on the translated AudioCaps and Clotho datasets show that our scheme achieves state-of-the-art performance on recall and consistency metrics for eight mainstream languages, including English. Our code will be available at https://github.com/ATRI-ACL/ATRI-ACL.
2502.14628
PEARL: Towards Permutation-Resilient LLMs
cs.LG cs.CL
The in-context learning (ICL) capability of large language models (LLMs) enables them to perform challenging tasks using provided demonstrations. However, ICL is highly sensitive to the ordering of demonstrations, leading to instability in predictions. This paper shows that this vulnerability can be exploited to design a natural attack - difficult for model providers to detect - that achieves nearly 80% success rate on LLaMA-3 by simply permuting the demonstrations. Existing mitigation methods primarily rely on post-processing and fail to enhance the model's inherent robustness to input permutations, raising concerns about safety and reliability of LLMs. To address this issue, we propose Permutation-resilient learning (PEARL), a novel framework based on distributionally robust optimization (DRO), which optimizes model performance against the worst-case input permutation. Specifically, PEARL consists of a permutation-proposal network (P-Net) and the LLM. The P-Net generates the most challenging permutations by treating it as an optimal transport problem, which is solved using an entropy-constrained Sinkhorn algorithm. Through minimax optimization, the P-Net and the LLM iteratively optimize against each other, progressively improving the LLM's robustness. Experiments on synthetic pre-training and real-world instruction tuning tasks demonstrate that PEARL effectively mitigates permutation attacks and enhances performance. Notably, despite being trained on fewer shots and shorter contexts, PEARL achieves performance gains of up to 40% when scaled to many-shot and long-context scenarios, highlighting its efficiency and generalization capabilities.
2502.14630
Understanding long-term energy use in off-grid solar home systems in sub-Saharan Africa
eess.SY cs.SY
Solar home systems provide low-cost electricity access for rural off-grid communities. As access to them increases, more long-term data becomes available on how these systems are used throughout their lifetime. This work analyses a dataset of 1,000 systems across sub-Saharan Africa. Dynamic time warping clustering was applied to the load demand data from the systems, identifying five distinct archetypal daily load profiles and their occurrence across the dataset. Temporal analysis reveals a general decline in daily energy consumption over time, with 57% of households reducing their usage after the first year of ownership. On average, there is a 33% decrease in daily consumption by the end of the second year compared to the peak demand, which occurs on the 96th day. Combining the load demand analysis with payment data shows that this decrease in energy consumption is observed even in households that are not experiencing economic hardship, indicating there are reasons beyond financial constraints for decreasing energy use once energy access is obtained.
2502.14631
Synergistic Fusion of Multi-Source Knowledge via Evidence Theory for High-Entropy Alloy Discovery
cs.LG
Discovering novel high-entropy alloys (HEAs) with desirable properties is challenging due to the vast compositional space and complex phase formation mechanisms. Efficient exploration of this space requires a strategic approach that integrates heterogeneous knowledge sources. Here, we propose a framework that systematically combines knowledge extracted from computational material datasets with domain knowledge distilled from scientific literature using large language models (LLMs). A central feature of this approach is the explicit consideration of element substitutability, identifying chemically similar elements that can be interchanged to potentially stabilize desired HEAs. Dempster-Shafer theory, a mathematical framework for reasoning under uncertainty, is employed to model and combine substitutabilities based on aggregated evidence from multiple sources. The framework predicts the phase stability of candidate HEA compositions and is systematically evaluated on both quaternary alloy systems, demonstrating superior performance compared to baseline machine learning models and methods reliant on single-source evidence in cross-validation experiments. By leveraging multi-source knowledge, the framework retains robust predictive power even when key elements are absent from the training data, underscoring its potential for knowledge transfer and extrapolation. Furthermore, the enhanced interpretability of the methodology offers insights into the fundamental factors governing HEA formation. Overall, this work provides a promising strategy for accelerating HEA discovery by integrating computational and textual knowledge sources, enabling efficient exploration of vast compositional spaces with improved generalization and interpretability.
2502.14634
CER: Confidence Enhanced Reasoning in LLMs
cs.LG
Ensuring the reliability of Large Language Models (LLMs) in complex reasoning tasks remains a formidable challenge, particularly in scenarios that demand precise mathematical calculations and knowledge-intensive open-domain generation. In this work, we introduce an uncertainty-aware framework designed to enhance the accuracy of LLM responses by systematically incorporating model confidence at critical decision points. We propose an approach that encourages multi-step reasoning in LLMs and quantify the confidence of intermediate answers such as numerical results in mathematical reasoning and proper nouns in open-domain generation. Then, the overall confidence of each reasoning chain is evaluated based on confidence of these critical intermediate steps. Finally, we aggregate the answer of generated response paths in a way that reflects the reliability of each generated content (as opposed to self-consistency in which each generated chain contributes equally to majority voting). We conducted extensive experiments in five datasets, three mathematical datasets and two open-domain datasets, using four LLMs. The results consistently validate the effectiveness of our novel confidence aggregation method, leading to an accuracy improvement of up to 7.4% and 5.8% over baseline approaches in math and open-domain generation tasks, respectively. Code is publicly available at https://github.com/ Aquasar11/CER.
2502.14637
ReQFlow: Rectified Quaternion Flow for Efficient and High-Quality Protein Backbone Generation
cs.LG cs.AI
Protein backbone generation plays a central role in de novo protein design and is significant for many biological and medical applications. Although diffusion and flow-based generative models provide potential solutions to this challenging task, they often generate proteins with undesired designability and suffer computational inefficiency. In this study, we propose a novel rectified quaternion flow (ReQFlow) matching method for fast and high-quality protein backbone generation. In particular, our method generates a local translation and a 3D rotation from random noise for each residue in a protein chain, which represents each 3D rotation as a unit quaternion and constructs its flow by spherical linear interpolation (SLERP) in an exponential format. We train the model by quaternion flow (QFlow) matching with guaranteed numerical stability and rectify the QFlow model to accelerate its inference and improve the designability of generated protein backbones, leading to the proposed ReQFlow model. Experiments show that ReQFlow achieves state-of-the-art performance in protein backbone generation while requiring much fewer sampling steps and significantly less inference time (e.g., being 37x faster than RFDiffusion and 62x faster than Genie2 when generating a backbone of length 300), demonstrating its effectiveness and efficiency. The code is available at https://github.com/AngxiaoYue/ReQFlow.
2502.14638
NAVIG: Natural Language-guided Analysis with Vision Language Models for Image Geo-localization
cs.CL cs.CV
Image geo-localization is the task of predicting the specific location of an image and requires complex reasoning across visual, geographical, and cultural contexts. While prior Vision Language Models (VLMs) have the best accuracy at this task, there is a dearth of high-quality datasets and models for analytical reasoning. We first create NaviClues, a high-quality dataset derived from GeoGuessr, a popular geography game, to supply examples of expert reasoning from language. Using this dataset, we present Navig, a comprehensive image geo-localization framework integrating global and fine-grained image information. By reasoning with language, Navig reduces the average distance error by 14% compared to previous state-of-the-art models while requiring fewer than 1000 training samples. Our dataset and code are available at https://github.com/SparrowZheyuan18/Navig/.
2502.14642
How Far are LLMs from Being Our Digital Twins? A Benchmark for Persona-Based Behavior Chain Simulation
cs.CL
Recently, LLMs have garnered increasing attention across academic disciplines for their potential as human digital twins, virtual proxies designed to replicate individuals and autonomously perform tasks such as decision-making, problem-solving, and reasoning on their behalf. However, current evaluations of LLMs primarily emphasize dialogue simulation while overlooking human behavior simulation, which is crucial for digital twins. To address this gap, we introduce BehaviorChain, the first benchmark for evaluating LLMs' ability to simulate continuous human behavior. BehaviorChain comprises diverse, high-quality, persona-based behavior chains, totaling 15,846 distinct behaviors across 1,001 unique personas, each with detailed history and profile metadata. For evaluation, we integrate persona metadata into LLMs and employ them to iteratively infer contextually appropriate behaviors within dynamic scenarios provided by BehaviorChain. Comprehensive evaluation results demonstrated that even state-of-the-art models struggle with accurately simulating continuous human behavior.
2502.14643
Length-Controlled Margin-Based Preference Optimization without Reference Model
cs.CL
Direct Preference Optimization (DPO) is a widely adopted offline algorithm for preference-based reinforcement learning from human feedback (RLHF), designed to improve training simplicity and stability by redefining reward functions. However, DPO is hindered by several limitations, including length bias, memory inefficiency, and probability degradation. To address these challenges, we propose Length-Controlled Margin-Based Preference Optimization (LMPO), a more efficient and robust alternative. LMPO introduces a uniform reference model as an upper bound for the DPO loss, enabling a more accurate approximation of the original optimization objective. Additionally, an average log-probability optimization strategy is employed to minimize discrepancies between training and inference phases. A key innovation of LMPO lies in its Length-Controlled Margin-Based loss function, integrated within the Bradley-Terry framework. This loss function regulates response length while simultaneously widening the margin between preferred and rejected outputs. By doing so, it mitigates probability degradation for both accepted and discarded responses, addressing a significant limitation of existing methods. We evaluate LMPO against state-of-the-art preference optimization techniques on two open-ended large language models, Mistral and LLaMA3, across six conditional benchmarks. Our experimental results demonstrate that LMPO effectively controls response length, reduces probability degradation, and outperforms existing approaches. The code is available at \url{https://github.com/gengxuli/LMPO}.
2502.14644
LIFT: Improving Long Context Understanding of Large Language Models through Long Input Fine-Tuning
cs.CL
Long context understanding remains challenging for large language models due to their limited context windows. This paper presents Long Input Fine-Tuning (LIFT), a novel framework for long-context modeling that can improve the long-context performance of arbitrary (short-context) LLMs by dynamically adapting model parameters based on the long input. Importantly, LIFT, rather than endlessly extending the context window size to accommodate increasingly longer inputs in context, chooses to store and absorb the long input in parameter. By fine-tuning the long input into model parameters, LIFT allows short-context LLMs to answer questions even when the required information is not provided in the context during inference. Furthermore, to enhance LIFT performance while maintaining the original in-context learning (ICL) capabilities, we introduce Gated Memory, a specialized attention adapter that automatically balances long input memorization and ICL. We provide a comprehensive analysis of the strengths and limitations of LIFT on long context understanding, offering valuable directions for future research.
2502.14645
Edit Once, Update Everywhere: A Simple Framework for Cross-Lingual Knowledge Synchronization in LLMs
cs.CL cs.AI
Knowledge editing allows for efficient adaptation of large language models (LLMs) to new information or corrections without requiring full retraining. However, prior methods typically focus on either single-language editing or basic multilingual editing, failing to achieve true cross-linguistic knowledge synchronization. To address this, we present a simple and practical state-of-the-art (SOTA) recipe Cross-Lingual Knowledge Democracy Edit (X-KDE), designed to propagate knowledge from a dominant language to other languages effectively. Our X-KDE comprises two stages: (i) Cross-lingual Edition Instruction Tuning (XE-IT), which fine-tunes the model on a curated parallel dataset to modify in-scope knowledge while preserving unrelated information, and (ii) Target-language Preference Optimization (TL-PO), which applies advanced optimization techniques to ensure consistency across languages, fostering the transfer of updates. Additionally, we contribute a high-quality, cross-lingual dataset, specifically designed to enhance knowledge transfer across languages. Extensive experiments on the Bi-ZsRE and MzsRE benchmarks show that X-KDE significantly enhances cross-lingual performance, achieving an average improvement of +8.19%, while maintaining high accuracy in monolingual settings.
2502.14648
Variance Reduction Methods Do Not Need to Compute Full Gradients: Improved Efficiency through Shuffling
cs.LG math.OC
In today's world, machine learning is hard to imagine without large training datasets and models. This has led to the use of stochastic methods for training, such as stochastic gradient descent (SGD). SGD provides weak theoretical guarantees of convergence, but there are modifications, such as Stochastic Variance Reduced Gradient (SVRG) and StochAstic Recursive grAdient algoritHm (SARAH), that can reduce the variance. These methods require the computation of the full gradient occasionally, which can be time consuming. In this paper, we explore variants of variance reduction algorithms that eliminate the need for full gradient computations. To make our approach memory-efficient and avoid full gradient computations, we use two key techniques: the shuffling heuristic and idea of SAG/SAGA methods. As a result, we improve existing estimates for variance reduction algorithms without the full gradient computations. Additionally, for the non-convex objective function, our estimate matches that of classic shuffling methods, while for the strongly convex one, it is an improvement. We conduct comprehensive theoretical analysis and provide extensive experimental results to validate the efficiency and practicality of our methods for large-scale machine learning problems.
2502.14659
MAGO-SP: Detection and Correction of Water-Fat Swaps in Magnitude-Only VIBE MRI
cs.CV
Volume Interpolated Breath-Hold Examination (VIBE) MRI generates images suitable for water and fat signal composition estimation. While the two-point VIBE provides water-fat-separated images, the six-point VIBE allows estimation of the effective transversal relaxation rate R2* and the proton density fat fraction (PDFF), which are imaging markers for health and disease. Ambiguity during signal reconstruction can lead to water-fat swaps. This shortcoming challenges the application of VIBE-MRI for automated PDFF analyses of large-scale clinical data and of population studies. This study develops an automated pipeline to detect and correct water-fat swaps in non-contrast-enhanced VIBE images. Our three-step pipeline begins with training a segmentation network to classify volumes as "fat-like" or "water-like," using synthetic water-fat swaps generated by merging fat and water volumes with Perlin noise. Next, a denoising diffusion image-to-image network predicts water volumes as signal priors for correction. Finally, we integrate this prior into a physics-constrained model to recover accurate water and fat signals. Our approach achieves a < 1% error rate in water-fat swap detection for a 6-point VIBE. Notably, swaps disproportionately affect individuals in the Underweight and Class 3 Obesity BMI categories. Our correction algorithm ensures accurate solution selection in chemical phase MRIs, enabling reliable PDFF estimation. This forms a solid technical foundation for automated large-scale population imaging analysis.
2502.14660
Beyond the Surface: Uncovering Implicit Locations with LLMs for Personalized Local News
cs.LG
News recommendation systems personalize homepage content to boost engagement, but factors like content type, editorial stance, and geographic focus impact recommendations. Local newspapers balance coverage across regions, yet identifying local articles is challenging due to implicit location cues like slang or landmarks. Traditional methods, such as Named Entity Recognition (NER) and Knowledge Graphs, infer locations, but Large Language Models (LLMs) offer new possibilities while raising concerns about accuracy and explainability. This paper explores LLMs for local article classification in Taboola's "Homepage For You" system, comparing them to traditional techniques. Key findings: (1) Knowledge Graphs enhance NER models' ability to detect implicit locations, (2) LLMs outperform traditional methods, and (3) LLMs can effectively identify local content without requiring Knowledge Graph integration. Offline evaluations showed LLMs excel at implicit location classification, while online A/B tests showed a significant increased in local views. A scalable pipeline integrating LLM-based location classification boosted local article distribution by 27%, preserving newspapers' brand identity and enhancing homepage personalization.
2502.14662
InstructAgent: Building User Controllable Recommender via LLM Agent
cs.CL cs.IR
Traditional recommender systems usually take the user-platform paradigm, where users are directly exposed under the control of the platform's recommendation algorithms. However, the defect of recommendation algorithms may put users in very vulnerable positions under this paradigm. First, many sophisticated models are often designed with commercial objectives in mind, focusing on the platform's benefits, which may hinder their ability to protect and capture users' true interests. Second, these models are typically optimized using data from all users, which may overlook individual user's preferences. Due to these shortcomings, users may experience several disadvantages under the traditional user-platform direct exposure paradigm, such as lack of control over the recommender system, potential manipulation by the platform, echo chamber effects, or lack of personalization for less active users due to the dominance of active users during collaborative learning. Therefore, there is an urgent need to develop a new paradigm to protect user interests and alleviate these issues. Recently, some researchers have introduced LLM agents to simulate user behaviors, these approaches primarily aim to optimize platform-side performance, leaving core issues in recommender systems unresolved. To address these limitations, we propose a new user-agent-platform paradigm, where agent serves as the protective shield between user and recommender system that enables indirect exposure. To this end, we first construct four recommendation datasets, denoted as $\dataset$, along with user instructions for each record.
2502.14663
The Restricted Isometry Property for Measurements from Group Orbits
cs.IT math.IT
It is known that sparse recovery by measurements from random circulant matrices provides good recovery bounds. We generalize this to measurements that arise as a random orbit of a group representation for some finite group G. We derive estimates for the number of measurements required to guarantee the restricted isometry property with high probability. Following this, we present several examples highlighting the role of appropriate representation-theoretic assumptions.
2502.14669
AlphaMaze: Enhancing Large Language Models' Spatial Intelligence via GRPO
cs.CL
Large Language Models (LLMs) have demonstrated impressive capabilities in language processing, yet they often struggle with tasks requiring genuine visual spatial reasoning. In this paper, we introduce a novel two-stage training framework designed to equip standard LLMs with visual reasoning abilities for maze navigation. First, we leverage Supervised Fine Tuning (SFT) on a curated dataset of tokenized maze representations to teach the model to predict step-by-step movement commands. Next, we apply Group Relative Policy Optimization (GRPO)-a technique used in DeepSeekR1-with a carefully crafted reward function to refine the model's sequential decision-making and encourage emergent chain-of-thought behaviors. Experimental results on synthetically generated mazes show that while a baseline model fails to navigate the maze, the SFT-trained model achieves 86% accuracy, and further GRPO fine-tuning boosts accuracy to 93%. Qualitative analyses reveal that GRPO fosters more robust and self-corrective reasoning, highlighting the potential of our approach to bridge the gap between language models and visual spatial tasks. These findings offer promising implications for applications in robotics, autonomous navigation, and other domains that require integrated visual and sequential reasoning.
2502.14671
Explanations of Deep Language Models Explain Language Representations in the Brain
cs.CL cs.AI q-bio.NC
Recent advances in artificial intelligence have given rise to large language models (LLMs) that not only achieve human-like performance but also share computational principles with the brain's language processing mechanisms. While previous research has primarily focused on aligning LLMs' internal representations with neural activity, we introduce a novel approach that leverages explainable AI (XAI) methods to forge deeper connections between the two domains. Using attribution methods, we quantified how preceding words contribute to an LLM's next-word predictions and employed these explanations to predict fMRI recordings from participants listening to the same narratives. Our findings demonstrate that attribution methods robustly predict brain activity across the language network, surpassing traditional internal representations in early language areas. This alignment is hierarchical: early-layer explanations correspond to the initial stages of language processing in the brain, while later layers align with more advanced stages. Moreover, the layers more influential on LLM next-word prediction$\unicode{x2014}$those with higher attribution scores$\unicode{x2014}$exhibited stronger alignment with neural activity. This work establishes a bidirectional bridge between AI and neuroscience. First, we demonstrate that attribution methods offer a powerful lens for investigating the neural mechanisms of language comprehension, revealing how meaning emerges from preceding context. Second, we propose using brain alignment as a metric to evaluate the validity of attribution methods, providing a framework for assessing their biological plausibility.
2502.14676
BP-SGCN: Behavioral Pseudo-Label Informed Sparse Graph Convolution Network for Pedestrian and Heterogeneous Trajectory Prediction
cs.CV cs.AI
Trajectory prediction allows better decision-making in applications of autonomous vehicles or surveillance by predicting the short-term future movement of traffic agents. It is classified into pedestrian or heterogeneous trajectory prediction. The former exploits the relatively consistent behavior of pedestrians, but is limited in real-world scenarios with heterogeneous traffic agents such as cyclists and vehicles. The latter typically relies on extra class label information to distinguish the heterogeneous agents, but such labels are costly to annotate and cannot be generalized to represent different behaviors within the same class of agents. In this work, we introduce the behavioral pseudo-labels that effectively capture the behavior distributions of pedestrians and heterogeneous agents solely based on their motion features, significantly improving the accuracy of trajectory prediction. To implement the framework, we propose the Behavioral Pseudo-Label Informed Sparse Graph Convolution Network (BP-SGCN) that learns pseudo-labels and informs to a trajectory predictor. For optimization, we propose a cascaded training scheme, in which we first learn the pseudo-labels in an unsupervised manner, and then perform end-to-end fine-tuning on the labels in the direction of increasing the trajectory prediction accuracy. Experiments show that our pseudo-labels effectively model different behavior clusters and improve trajectory prediction. Our proposed BP-SGCN outperforms existing methods using both pedestrian (ETH/UCY, pedestrian-only SDD) and heterogeneous agent datasets (SDD, Argoverse 1).
2502.14677
Data-Constrained Synthesis of Training Data for De-Identification
cs.CL cs.AI
Many sensitive domains -- such as the clinical domain -- lack widely available datasets due to privacy risks. The increasing generative capabilities of large language models (LLMs) have made synthetic datasets a viable path forward. In this study, we domain-adapt LLMs to the clinical domain and generate synthetic clinical texts that are machine-annotated with tags for personally identifiable information using capable encoder-based NER models. The synthetic corpora are then used to train synthetic NER models. The results show that training NER models using synthetic corpora incurs only a small drop in predictive performance. The limits of this process are investigated in a systematic ablation study -- using both Swedish and Spanish data. Our analysis shows that smaller datasets can be sufficient for domain-adapting LLMs for data synthesis. Instead, the effectiveness of this process is almost entirely contingent on the performance of the machine-annotating NER models trained using the original data.
2502.14678
How to Get Your LLM to Generate Challenging Problems for Evaluation
cs.CL
The pace of evolution of Large Language Models (LLMs) necessitates new approaches for rigorous and comprehensive evaluation. Traditional human annotation is increasingly impracticable due to the complexities and costs involved in generating high-quality, challenging problems. In this work, we introduce CHASE, a unified framework to synthetically generate challenging problems using LLMs without human involvement. For a given task, our approach builds a hard problem in a bottom-up manner from simpler components. Moreover, our framework decomposes the generation process into independently verifiable sub-tasks, thereby ensuring a high level of quality and correctness. We implement CHASE to create evaluation benchmarks across three diverse domains: (1) document-based question answering, (2) repository-level code completion, and (3) math reasoning. The performance of state-of-the-art LLMs on these synthetic benchmarks lies in the range of 40-60% accuracy, thereby demonstrating the effectiveness of our framework at generating challenging problems. We publicly release our benchmarks and code.
2502.14679
Disentangled Latent Spaces for Reduced Order Models using Deterministic Autoencoders
cs.LG
Data-driven reduced-order models based on autoencoders generally lack interpretability compared to classical methods such as the proper orthogonal decomposition. More interpretability can be gained by disentangling the latent variables and analyzing the resulting modes. For this purpose, probabilistic $\beta$-variational autoencoders ($\beta$-VAEs) are frequently used in computational fluid dynamics and other simulation sciences. Using a benchmark periodic flow dataset, we show that competitive results can be achieved using non-probabilistic autoencoder approaches that either promote orthogonality or penalize correlation between latent variables. Compared to probabilistic autoencoders, these approaches offer more robustness with respect to the choice of hyperparameters entering the loss function. We further demonstrate the ability of a non-probabilistic approach to identify a reduced number of active latent variables by introducing a correlation penalty, a function also known from the use of $\beta$-VAE. The investigated probabilistic and non-probabilistic autoencoder models are finally used for the dimensionality reduction of aircraft ditching loads, which serves as an industrial application in this work.
2502.14681
seqKAN: Sequence processing with Kolmogorov-Arnold Networks
cs.LG cs.AI
Kolmogorov-Arnold Networks (KANs) have been recently proposed as a machine learning framework that is more interpretable and controllable than the multi-layer perceptron. Various network architectures have been proposed within the KAN framework targeting different tasks and application domains, including sequence processing. This paper proposes seqKAN, a new KAN architecture for sequence processing. Although multiple sequence processing KAN architectures have already been proposed, we argue that seqKAN is more faithful to the core concept of the KAN framework. Furthermore, we empirically demonstrate that it achieves better results. The empirical evaluation is performed on generated data from a complex physics problem on an interpolation and an extrapolation task. Using this dataset we compared seqKAN against a prior KAN network for timeseries prediction, recurrent deep networks, and symbolic regression. seqKAN substantially outperforms all architectures, particularly on the extrapolation dataset, while also being the most transparent.
2502.14682
Bridging the Gap: Transforming Natural Language Questions into SQL Queries via Abstract Query Pattern and Contextual Schema Markup
cs.CL
Large language models have demonstrated excellent performance in many tasks, including Text-to-SQL, due to their powerful in-context learning capabilities. They are becoming the mainstream approach for Text-to-SQL. However, these methods still have a significant gap compared to human performance, especially on complex questions. As the complexity of questions increases, the gap between questions and SQLs increases. We identify two important gaps: the structural mapping gap and the lexical mapping gap. To tackle these two gaps, we propose PAS-SQL, an efficient SQL generation pipeline based on LLMs, which alleviates gaps through Abstract Query Pattern (AQP) and Contextual Schema Markup (CSM). AQP aims to obtain the structural pattern of the question by removing database-related information, which enables us to find structurally similar demonstrations. CSM aims to associate database-related text span in the question with specific tables or columns in the database, which alleviates the lexical mapping gap. Experimental results on the Spider and BIRD datasets demonstrate the effectiveness of our proposed method. Specifically, PAS-SQL + GPT-4o sets a new state-of-the-art on the Spider benchmark with an execution accuracy of 87.9\%, and achieves leading results on the BIRD dataset with an execution accuracy of 64.67\%.
2502.14684
CDGS: Confidence-Aware Depth Regularization for 3D Gaussian Splatting
cs.GR cs.CV
3D Gaussian Splatting (3DGS) has shown significant advantages in novel view synthesis (NVS), particularly in achieving high rendering speeds and high-quality results. However, its geometric accuracy in 3D reconstruction remains limited due to the lack of explicit geometric constraints during optimization. This paper introduces CDGS, a confidence-aware depth regularization approach developed to enhance 3DGS. We leverage multi-cue confidence maps of monocular depth estimation and sparse Structure-from-Motion depth to adaptively adjust depth supervision during the optimization process. Our method demonstrates improved geometric detail preservation in early training stages and achieves competitive performance in both NVS quality and geometric accuracy. Experiments on the publicly available Tanks and Temples benchmark dataset show that our method achieves more stable convergence behavior and more accurate geometric reconstruction results, with improvements of up to 2.31 dB in PSNR for NVS and consistently lower geometric errors in M3C2 distance metrics. Notably, our method reaches comparable F-scores to the original 3DGS with only 50% of the training iterations. We expect this work will facilitate the development of efficient and accurate 3D reconstruction systems for real-world applications such as digital twin creation, heritage preservation, or forestry applications.
2502.14689
Confidence Estimation via Sequential Likelihood Mixing
stat.ML cs.LG
We present a universal framework for constructing confidence sets based on sequential likelihood mixing. Building upon classical results from sequential analysis, we provide a unifying perspective on several recent lines of work, and establish fundamental connections between sequential mixing, Bayesian inference and regret inequalities from online estimation. The framework applies to any realizable family of likelihood functions and allows for non-i.i.d. data and anytime validity. Moreover, the framework seamlessly integrates standard approximate inference techniques, such as variational inference and sampling-based methods, and extends to misspecified model classes, while preserving provable coverage guarantees. We illustrate the power of the framework by deriving tighter confidence sequences for classical settings, including sequential linear regression and sparse estimation, with simplified proofs.
2502.14693
I-MCTS: Enhancing Agentic AutoML via Introspective Monte Carlo Tree Search
cs.CL
Recent advancements in large language models (LLMs) have shown remarkable potential in automating machine learning tasks. However, existing LLM-based agents often struggle with low-diversity and suboptimal code generation. While recent work has introduced Monte Carlo Tree Search (MCTS) to address these issues, limitations persist in the quality and diversity of thoughts generated, as well as in the scalar value feedback mechanisms used for node selection. In this study, we introduce Introspective Monte Carlo Tree Search (I-MCTS), a novel approach that iteratively expands tree nodes through an introspective process that meticulously analyzes solutions and results from parent and sibling nodes. This facilitates a continuous refinement of the node in the search tree, thereby enhancing the overall decision-making process.Furthermore, we integrate a Large Language Model (LLM)-based value model to facilitate direct evaluation of each node's solution prior to conducting comprehensive computational rollouts. A hybrid rewarding mechanism is implemented to seamlessly transition the Q-value from LLM-estimated scores to actual performance scores. This allows higher-quality nodes to be traversed earlier.Applied to the various ML tasks, our approach demonstrates a6\% absolute improvement in performance compared to the strong open-source AutoML agents, showcasing its effectiveness in enhancing agentic AutoML systems.
2502.14694
Revisiting Near-Far Field Boundary in Dual-Polarized XL-MIMO Systems
cs.IT math.IT
Extremely large-scale multiple-input multiple-output (XL-MIMO) is expected to be an important technology in future sixth generation (6G) networks. Compared with conventional single-polarized XL-MIMO, where signals are transmitted and received in only one polarization direction, dual-polarized XL-MIMO systems achieve higher data rate by improving multiplexing performances, and thus are the focus of this paper. Due to enlarged aperture, near-field regions become non-negligible in XL-MIMO communications, necessitating accurate near-far field boundary characterizations. However, existing boundaries developed for single-polarized systems only consider phase or power differences across array elements while irrespective of cross-polarization discrimination (XPD) variances in dual-polarized XL-MIMO systems, deteriorating transmit covariance optimization performances. In this paper, we revisit near-far field boundaries for dual-polarized XL-MIMO systems by taking XPD differences into account, which faces the following challenge. Unlike existing near-far field boundaries, which only need to consider co-polarized channel components, deriving boundaries for dual-polarized XL-MIMO systems requires modeling joint effects of co-polarized and cross-polarized components. To address this issue, we model XPD variations across antennas and introduce a non-uniform XPD distance to complement existing near-far field boundaries. Based on the new distance criterion, we propose an efficient scheme to optimize transmit covariance. Numerical results validate our analysis and demonstrate the proposed algorithm's effectiveness.
2502.14698
General Uncertainty Estimation with Delta Variances
cs.LG cs.AI stat.AP stat.ML
Decision makers may suffer from uncertainty induced by limited data. This may be mitigated by accounting for epistemic uncertainty, which is however challenging to estimate efficiently for large neural networks. To this extent we investigate Delta Variances, a family of algorithms for epistemic uncertainty quantification, that is computationally efficient and convenient to implement. It can be applied to neural networks and more general functions composed of neural networks. As an example we consider a weather simulator with a neural-network-based step function inside -- here Delta Variances empirically obtain competitive results at the cost of a single gradient computation. The approach is convenient as it requires no changes to the neural network architecture or training procedure. We discuss multiple ways to derive Delta Variances theoretically noting that special cases recover popular techniques and present a unified perspective on multiple related methods. Finally we observe that this general perspective gives rise to a natural extension and empirically show its benefit.
2502.14704
Not All Data are Good Labels: On the Self-supervised Labeling for Time Series Forecasting
cs.LG cs.AI
Time Series Forecasting (TSF) is a crucial task in various domains, yet existing TSF models rely heavily on high-quality data and insufficiently exploit all available data. This paper explores a novel self-supervised approach to re-label time series datasets by inherently constructing candidate datasets. During the optimization of a simple reconstruction network, intermediates are used as pseudo labels in a self-supervised paradigm, improving generalization for any predictor. We introduce the Self-Correction with Adaptive Mask (SCAM), which discards overfitted components and selectively replaces them with pseudo labels generated from reconstructions. Additionally, we incorporate Spectral Norm Regularization (SNR) to further suppress overfitting from a loss landscape perspective. Our experiments on eleven real-world datasets demonstrate that SCAM consistently improves the performance of various backbone models. This work offers a new perspective on constructing datasets and enhancing the generalization of TSF models through self-supervised learning.
2502.14706
Building reliable sim driving agents by scaling self-play
cs.AI cs.RO
Simulation agents are essential for designing and testing systems that interact with humans, such as autonomous vehicles (AVs). These agents serve various purposes, from benchmarking AV performance to stress-testing the system's limits, but all use cases share a key requirement: reliability. A simulation agent should behave as intended by the designer, minimizing unintended actions like collisions that can compromise the signal-to-noise ratio of analyses. As a foundation for reliable sim agents, we propose scaling self-play to thousands of scenarios on the Waymo Open Motion Dataset under semi-realistic limits on human perception and control. Training from scratch on a single GPU, our agents nearly solve the full training set within a day. They generalize effectively to unseen test scenes, achieving a 99.8% goal completion rate with less than 0.8% combined collision and off-road incidents across 10,000 held-out scenarios. Beyond in-distribution generalization, our agents show partial robustness to out-of-distribution scenes and can be fine-tuned in minutes to reach near-perfect performance in those cases. Demonstrations of agent behaviors can be found at this link. We open-source both the pre-trained agents and the complete code base. Demonstrations of agent behaviors can be found at \url{https://sites.google.com/view/reliable-sim-agents}.
2502.14707
TRUSWorthy: Toward Clinically Applicable Deep Learning for Confident Detection of Prostate Cancer in Micro-Ultrasound
eess.IV cs.LG q-bio.TO
While deep learning methods have shown great promise in improving the effectiveness of prostate cancer (PCa) diagnosis by detecting suspicious lesions from trans-rectal ultrasound (TRUS), they must overcome multiple simultaneous challenges. There is high heterogeneity in tissue appearance, significant class imbalance in favor of benign examples, and scarcity in the number and quality of ground truth annotations available to train models. Failure to address even a single one of these problems can result in unacceptable clinical outcomes.We propose TRUSWorthy, a carefully designed, tuned, and integrated system for reliable PCa detection. Our pipeline integrates self-supervised learning, multiple-instance learning aggregation using transformers, random-undersampled boosting and ensembling: these address label scarcity, weak labels, class imbalance, and overconfidence, respectively. We train and rigorously evaluate our method using a large, multi-center dataset of micro-ultrasound data. Our method outperforms previous state-of-the-art deep learning methods in terms of accuracy and uncertainty calibration, with AUROC and balanced accuracy scores of 79.9% and 71.5%, respectively. On the top 20% of predictions with the highest confidence, we can achieve a balanced accuracy of up to 91%. The success of TRUSWorthy demonstrates the potential of integrated deep learning solutions to meet clinical needs in a highly challenging deployment setting, and is a significant step towards creating a trustworthy system for computer-assisted PCa diagnosis.
2502.14708
Human Misperception of Generative-AI Alignment: A Laboratory Experiment
econ.TH cs.AI cs.GT
We conduct an incentivized laboratory experiment to study people's perception of generative artificial intelligence (GenAI) alignment in the context of economic decision-making. Using a panel of economic problems spanning the domains of risk, time preference, social preference, and strategic interactions, we ask human subjects to make choices for themselves and to predict the choices made by GenAI on behalf of a human user. We find that people overestimate the degree of alignment between GenAI's choices and human choices. In every problem, human subjects' average prediction about GenAI's choice is substantially closer to the average human-subject choice than it is to the GenAI choice. At the individual level, different subjects' predictions about GenAI's choice in a given problem are highly correlated with their own choices in the same problem. We explore the implications of people overestimating GenAI alignment in a simple theoretical model.
2502.14709
Data-Efficient Pretraining with Group-Level Data Influence Modeling
cs.CL cs.LG
Data-efficient pretraining has shown tremendous potential to elevate scaling laws. This paper argues that effective pretraining data should be curated at the group level, treating a set of data points as a whole rather than as independent contributors. To achieve that, we propose Group-Level Data Influence Modeling (Group-MATES), a novel data-efficient pretraining method that captures and optimizes group-level data utility. Specifically, Group-MATES collects oracle group-level influences by locally probing the pretraining model with data sets. It then fine-tunes a relational data influence model to approximate oracles as relationship-weighted aggregations of individual influences. The fine-tuned model selects the data subset by maximizing its group-level influence prediction, with influence-aware clustering to enable efficient inference. Experiments on the DCLM benchmark demonstrate that Group-MATES achieves a 10% relative core score improvement on 22 downstream tasks over DCLM-Baseline and 5% over individual-influence-based methods, establishing a new state-of-the-art. Further analyses highlight the effectiveness of relational data influence models in capturing intricate interactions between data points.
2502.14714
From Knowledge Generation to Knowledge Verification: Examining the BioMedical Generative Capabilities of ChatGPT
cs.AI cs.CL cs.IR
The generative capabilities of LLM models present opportunities in accelerating tasks and concerns with the authenticity of the knowledge it produces. To address the concerns, we present a computational approach that systematically evaluates the factual accuracy of biomedical knowledge that an LLM model has been prompted to generate. Our approach encompasses two processes: the generation of disease-centric associations and the verification of them using the semantic knowledge of the biomedical ontologies. Using ChatGPT as the select LLM model, we designed a set of prompt-engineering processes to generate linkages between diseases, drugs, symptoms, and genes to establish grounds for assessments. Experimental results demonstrate high accuracy in identifying disease terms (88%-97%), drug names (90%-91%), and genetic information (88%-98%). The symptom term identification accuracy was notably lower (49%-61%), as verified against the DOID, ChEBI, SYMPTOM, and GO ontologies accordingly. The verification of associations reveals literature coverage rates of (89%-91%) among disease-drug and disease-gene associations. The low identification accuracy for symptom terms also contributed to the verification of symptom-related associations (49%-62%).
2502.14718
Entity Framing and Role Portrayal in the News
cs.CL
We introduce a novel multilingual hierarchical corpus annotated for entity framing and role portrayal in news articles. The dataset uses a unique taxonomy inspired by storytelling elements, comprising 22 fine-grained roles, or archetypes, nested within three main categories: protagonist, antagonist, and innocent. Each archetype is carefully defined, capturing nuanced portrayals of entities such as guardian, martyr, and underdog for protagonists; tyrant, deceiver, and bigot for antagonists; and victim, scapegoat, and exploited for innocents. The dataset includes 1,378 recent news articles in five languages (Bulgarian, English, Hindi, European Portuguese, and Russian) focusing on two critical domains of global significance: the Ukraine-Russia War and Climate Change. Over 5,800 entity mentions have been annotated with role labels. This dataset serves as a valuable resource for research into role portrayal and has broader implications for news analysis. We describe the characteristics of the dataset and the annotation process, and we report evaluation results on fine-tuned state-of-the-art multilingual transformers and hierarchical zero-shot learning using LLMs at the level of a document, a paragraph, and a sentence.
2502.14719
Internal Incoherency Scores for Constraint-based Causal Discovery Algorithms
stat.ML cs.LG
Causal discovery aims to infer causal graphs from observational or experimental data. Methods such as the popular PC algorithm are based on conditional independence testing and utilize enabling assumptions, such as the faithfulness assumption, for their inferences. In practice, these assumptions, as well as the functional assumptions inherited from the chosen conditional independence test, are typically taken as a given and not further tested for their validity on the data. In this work, we propose internal coherency scores that allow testing for assumption violations and finite sample errors, whenever detectable without requiring ground truth or further statistical tests. We provide a complete classification of erroneous results, including a distinction between detectable and undetectable errors, and prove that the detectable erroneous results can be measured by our scores. We illustrate our coherency scores on the PC algorithm with simulated and real-world datasets, and envision that testing for internal coherency can become a standard tool in applying constraint-based methods, much like a suite of tests is used to validate the assumptions of classical regression analysis.
2502.14720
Advancing Measurement Capabilities in Lithium-Ion Batteries: Exploring the Potential of Fiber Optic Sensors for Thermal Monitoring of Battery Cells
physics.app-ph cs.SY eess.SY
This work demonstrates the potential of fiber optic sensors for measuring thermal effects in lithium-ion batteries, using a fiber optic measurement method of Optical Frequency Domain Reflectometry (OFDR). The innovative application of fiber sensors allows for spatially resolved temperature measurement, particularly emphasizing the importance of monitoring not just the exterior but also the internal conditions within battery cells. Utilizing inert glass fibers as sensors, which exhibit minimal sensitivity to electric fields, opens up new pathways for their implementation in a wide range of applications, such as battery monitoring. The sensors used in this work provide real-time information along the entire length of the fiber, unlike commonly used Fiber Bragg Grating (FBG) sensors. It is shown that using the herein presented novel sensors in a temperature range of 0 to 80 degree celsius reveals a linear thermal dependency with high sensitivity and a local resolution of a few centimeters. Furthermore, this study presents preliminary findings on the potential application of fiber optic sensors in lithium-ion battery (LIB) cells, demonstrating that the steps required for battery integration do not impose any restrictive effects on thermal measurements.
2502.14721
Multi-dataset synergistic in supervised learning to pre-label structural components in point clouds from shell construction scenes
cs.CV
The significant effort required to annotate data for new training datasets hinders computer vision research and machine learning in the construction industry. This work explores adapting standard datasets and the latest transformer model architectures for point cloud semantic segmentation in the context of shell construction sites. Unlike common approaches focused on object segmentation of building interiors and furniture, this study addressed the challenges of segmenting complex structural components in Architecture, Engineering, and Construction (AEC). We establish a baseline through supervised training and a custom validation dataset, evaluate the cross-domain inference with large-scale indoor datasets, and utilize transfer learning to maximize segmentation performance with minimal new data. The findings indicate that with minimal fine-tuning, pre-trained transformer architectures offer an effective strategy for building component segmentation. Our results are promising for automating the annotation of new, previously unseen data when creating larger training resources and for the segmentation of frequently recurring objects.
2502.14724
Ranking Joint Policies in Dynamic Games using Evolutionary Dynamics
cs.MA cs.AI cs.LG
Game-theoretic solution concepts, such as the Nash equilibrium, have been key to finding stable joint actions in multi-player games. However, it has been shown that the dynamics of agents' interactions, even in simple two-player games with few strategies, are incapable of reaching Nash equilibria, exhibiting complex and unpredictable behavior. Instead, evolutionary approaches can describe the long-term persistence of strategies and filter out transient ones, accounting for the long-term dynamics of agents' interactions. Our goal is to identify agents' joint strategies that result in stable behavior, being resistant to changes, while also accounting for agents' payoffs, in dynamic games. Towards this goal, and building on previous results, this paper proposes transforming dynamic games into their empirical forms by considering agents' strategies instead of agents' actions, and applying the evolutionary methodology $\alpha$-Rank to evaluate and rank strategy profiles according to their long-term dynamics. This methodology not only allows us to identify joint strategies that are strong through agents' long-term interactions, but also provides a descriptive, transparent framework regarding the high ranking of these strategies. Experiments report on agents that aim to collaboratively solve a stochastic version of the graph coloring problem. We consider different styles of play as strategies to define the empirical game, and train policies realizing these strategies, using the DQN algorithm. Then we run simulations to generate the payoff matrix required by $\alpha$-Rank to rank joint strategies.
2502.14727
WavRAG: Audio-Integrated Retrieval Augmented Generation for Spoken Dialogue Models
cs.SD cs.AI eess.AS
Retrieval Augmented Generation (RAG) has gained widespread adoption owing to its capacity to empower large language models (LLMs) to integrate external knowledge. However, existing RAG frameworks are primarily designed for text-based LLMs and rely on Automatic Speech Recognition to process speech input, which discards crucial audio information, risks transcription errors, and increases computational overhead. Therefore, we introduce WavRAG, the first retrieval augmented generation framework with native, end-to-end audio support. WavRAG offers two key features: 1) Bypassing ASR, WavRAG directly processes raw audio for both embedding and retrieval. 2) WavRAG integrates audio and text into a unified knowledge representation. Specifically, we propose the WavRetriever to facilitate the retrieval from a text-audio hybrid knowledge base, and further enhance the in-context capabilities of spoken dialogue models through the integration of chain-of-thought reasoning. In comparison to state-of-the-art ASR-Text RAG pipelines, WavRAG achieves comparable retrieval performance while delivering a 10x acceleration. Furthermore, WavRAG's unique text-audio hybrid retrieval capability extends the boundaries of RAG to the audio modality.
2502.14731
Beyond Performance Scores: Directed Functional Connectivity as a Brain-Based Biomarker for Motor Skill Learning and Retention
q-bio.NC cs.LG
Motor skill acquisition in fields like surgery, robotics, and sports involves learning complex task sequences through extensive training. Traditional performance metrics, like execution time and error rates, offer limited insight as they fail to capture the neural mechanisms underlying skill learning and retention. This study introduces directed functional connectivity (dFC), derived from electroencephalography (EEG), as a novel brain-based biomarker for assessing motor skill learning and retention. For the first time, dFC is applied as a biomarker to map the stages of the Fitts and Posner motor learning model, offering new insights into the neural mechanisms underlying skill acquisition and retention. Unlike traditional measures, it captures both the strength and direction of neural information flow, providing a comprehensive understanding of neural adaptations across different learning stages. The analysis demonstrates that dFC can effectively identify and track the progression through various stages of the Fitts and Posner model. Furthermore, its stability over a six-week washout period highlights its utility in monitoring long-term retention. No significant changes in dFC were observed in a control group, confirming that the observed neural adaptations were specific to training and not due to external factors. By offering a granular view of the learning process at the group and individual levels, dFC facilitates the development of personalized, targeted training protocols aimed at enhancing outcomes in fields where precision and long-term retention are critical, such as surgical education. These findings underscore the value of dFC as a robust biomarker that complements traditional performance metrics, providing a deeper understanding of motor skill learning and retention.
2502.14734
Sentence Smith: Formally Controllable Text Transformation and its Application to Evaluation of Text Embedding Models
cs.CL
We propose the Sentence Smith framework that enables controlled and specified manipulation of text meaning. It consists of three main steps: 1. Parsing a sentence into a semantic graph, 2. Applying human-designed semantic manipulation rules, and 3. Generating text from the manipulated graph. A final filtering step (4.) ensures the validity of the applied transformation. To demonstrate the utility of Sentence Smith in an application study, we use it to generate hard negative pairs that challenge text embedding models. Since the controllable generation makes it possible to clearly isolate different types of semantic shifts, we can gain deeper insights into the specific strengths and weaknesses of widely used text embedding models, also addressing an issue in current benchmarking where linguistic phenomena remain opaque. Human validation confirms that the generations produced by Sentence Smith are highly accurate.
2502.14735
EAGER-LLM: Enhancing Large Language Models as Recommenders through Exogenous Behavior-Semantic Integration
cs.IR cs.AI
Large language models (LLMs) are increasingly leveraged as foundational backbones in the development of advanced recommender systems, offering enhanced capabilities through their extensive knowledge and reasoning. Existing llm-based recommender systems (RSs) often face challenges due to the significant differences between the linguistic semantics of pre-trained LLMs and the collaborative semantics essential for RSs. These systems use pre-trained linguistic semantics but learn collaborative semantics from scratch via the llm-Backbone. However, LLMs are not designed for recommendations, leading to inefficient collaborative learning, weak result correlations, and poor integration of traditional RS features. To address these challenges, we propose EAGER-LLM, a decoder-only llm-based generative recommendation framework that integrates endogenous and exogenous behavioral and semantic information in a non-intrusive manner. Specifically, we propose 1)dual-source knowledge-rich item indices that integrates indexing sequences for exogenous signals, enabling efficient link-wide processing; 2)non-invasive multiscale alignment reconstruction tasks guide the model toward a deeper understanding of both collaborative and semantic signals; 3)an annealing adapter designed to finely balance the model's recommendation performance with its comprehension capabilities. We demonstrate EAGER-LLM's effectiveness through rigorous testing on three public benchmarks.
2502.14738
Robust Information Selection for Hypothesis Testing with Misclassification Penalties
stat.ML cs.SY eess.SP eess.SY math.CO math.OC
We study the problem of robust information selection for a Bayesian hypothesis testing / classification task, where the goal is to identify the true state of the world from a finite set of hypotheses based on observations from the selected information sources. We introduce a novel misclassification penalty framework, which enables non-uniform treatment of different misclassification events. Extending the classical subset selection framework, we study the problem of selecting a subset of sources that minimize the maximum penalty of misclassification under a limited budget, despite deletions or failures of a subset of the selected sources. We characterize the curvature properties of the objective function and propose an efficient greedy algorithm with performance guarantees. Next, we highlight certain limitations of optimizing for the maximum penalty metric and propose a submodular surrogate metric to guide the selection of the information set. We propose a greedy algorithm with near-optimality guarantees for optimizing the surrogate metric. Finally, we empirically demonstrate the performance of our proposed algorithms in several instances of the information set selection problem.
2502.14739
SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines
cs.CL
Large language models (LLMs) have demonstrated remarkable proficiency in mainstream academic disciplines such as mathematics, physics, and computer science. However, human knowledge encompasses over 200 specialized disciplines, far exceeding the scope of existing benchmarks. The capabilities of LLMs in many of these specialized fields-particularly in light industry, agriculture, and service-oriented disciplines-remain inadequately evaluated. To address this gap, we present SuperGPQA, a comprehensive benchmark that evaluates graduate-level knowledge and reasoning capabilities across 285 disciplines. Our benchmark employs a novel Human-LLM collaborative filtering mechanism to eliminate trivial or ambiguous questions through iterative refinement based on both LLM responses and expert feedback. Our experimental results reveal significant room for improvement in the performance of current state-of-the-art LLMs across diverse knowledge domains (e.g., the reasoning-focused model DeepSeek-R1 achieved the highest accuracy of 61.82% on SuperGPQA), highlighting the considerable gap between current model capabilities and artificial general intelligence. Additionally, we present comprehensive insights from our management of a large-scale annotation process, involving over 80 expert annotators and an interactive Human-LLM collaborative system, offering valuable methodological guidance for future research initiatives of comparable scope.
2502.14740
YOLOv12: A Breakdown of the Key Architectural Features
cs.CV cs.AI
This paper presents an architectural analysis of YOLOv12, a significant advancement in single-stage, real-time object detection building upon the strengths of its predecessors while introducing key improvements. The model incorporates an optimised backbone (R-ELAN), 7x7 separable convolutions, and FlashAttention-driven area-based attention, improving feature extraction, enhanced efficiency, and robust detections. With multiple model variants, similar to its predecessors, YOLOv12 offers scalable solutions for both latency-sensitive and high-accuracy applications. Experimental results manifest consistent gains in mean average precision (mAP) and inference speed, making YOLOv12 a compelling choice for applications in autonomous systems, security, and real-time analytics. By achieving an optimal balance between computational efficiency and performance, YOLOv12 sets a new benchmark for real-time computer vision, facilitating deployment across diverse hardware platforms, from edge devices to high-performance clusters.
2502.14741
Reinforcement Learning with Graph Attention for Routing and Wavelength Assignment with Lightpath Reuse
cs.NI cs.LG cs.SY eess.SY
Many works have investigated reinforcement learning (RL) for routing and spectrum assignment on flex-grid networks but only one work to date has examined RL for fixed-grid with flex-rate transponders, despite production systems using this paradigm. Flex-rate transponders allow existing lightpaths to accommodate new services, a task we term routing and wavelength assignment with lightpath reuse (RWA-LR). We re-examine this problem and present a thorough benchmarking of heuristic algorithms for RWA-LR, which are shown to have 6% increased throughput when candidate paths are ordered by number of hops, rather than total length. We train an RL agent for RWA-LR with graph attention networks for the policy and value functions to exploit the graph-structured data. We provide details of our methodology and open source all of our code for reproduction. We outperform the previous state-of-the-art RL approach by 2.5% (17.4 Tbps mean additional throughput) and the best heuristic by 1.2% (8.5 Tbps mean additional throughput). This marginal gain highlights the difficulty in learning effective RL policies on long horizon resource allocation tasks.
2502.14743
Multi-Agent Coordination across Diverse Applications: A Survey
cs.MA cs.AI
Multi-agent coordination studies the underlying mechanism enabling the trending spread of diverse multi-agent systems (MAS) and has received increasing attention, driven by the expansion of emerging applications and rapid AI advances. This survey outlines the current state of coordination research across applications through a unified understanding that answers four fundamental coordination questions: (1) what is coordination; (2) why coordination; (3) who to coordinate with; and (4) how to coordinate. Our purpose is to explore existing ideas and expertise in coordination and their connections across diverse applications, while identifying and highlighting emerging and promising research directions. First, general coordination problems that are essential to varied applications are identified and analyzed. Second, a number of MAS applications are surveyed, ranging from widely studied domains, e.g., search and rescue, warehouse automation and logistics, and transportation systems, to emerging fields including humanoid and anthropomorphic robots, satellite systems, and large language models (LLMs). Finally, open challenges about the scalability, heterogeneity, and learning mechanisms of MAS are analyzed and discussed. In particular, we identify the hybridization of hierarchical and decentralized coordination, human-MAS coordination, and LLM-based MAS as promising future directions.
2502.14744
HiddenDetect: Detecting Jailbreak Attacks against Large Vision-Language Models via Monitoring Hidden States
cs.CL
The integration of additional modalities increases the susceptibility of large vision-language models (LVLMs) to safety risks, such as jailbreak attacks, compared to their language-only counterparts. While existing research primarily focuses on post-hoc alignment techniques, the underlying safety mechanisms within LVLMs remain largely unexplored. In this work , we investigate whether LVLMs inherently encode safety-relevant signals within their internal activations during inference. Our findings reveal that LVLMs exhibit distinct activation patterns when processing unsafe prompts, which can be leveraged to detect and mitigate adversarial inputs without requiring extensive fine-tuning. Building on this insight, we introduce HiddenDetect, a novel tuning-free framework that harnesses internal model activations to enhance safety. Experimental results show that {HiddenDetect} surpasses state-of-the-art methods in detecting jailbreak attacks against LVLMs. By utilizing intrinsic safety-aware patterns, our method provides an efficient and scalable solution for strengthening LVLM robustness against multimodal threats. Our code will be released publicly at https://github.com/leigest519/HiddenDetect.
2502.14745
SQL4NN: Validation and expressive querying of models as data
cs.DB cs.LG
We consider machine learning models, learned from data, to be an important, intensional, kind of data in themselves. As such, various analysis tasks on models can be thought of as queries over this intensional data, often combined with extensional data such as data for training or validation. We demonstrate that relational database systems and SQL can actually be well suited for many such tasks.
2502.14746
Classical and quantum Coxeter codes: Extending the Reed-Muller family
cs.IT math.CO math.IT quant-ph
We introduce a class of binary linear codes that generalizes the Reed-Muller family by replacing the group $\mathbb{Z}_2^m$ with an arbitrary finite Coxeter group. Similar to the Reed-Muller codes, this class is closed under duality and has rate determined by a Gaussian distribution. We also construct quantum CSS codes arising from the Coxeter codes, which admit transversal logical operators outside of the Clifford group.
2502.14748
Large Language Models Struggle to Describe the Haystack without Human Help: Human-in-the-loop Evaluation of LLMs
cs.CL
A common use of NLP is to facilitate the understanding of large document collections, with a shift from using traditional topic models to Large Language Models. Yet the effectiveness of using LLM for large corpus understanding in real-world applications remains under-explored. This study measures the knowledge users acquire with unsupervised, supervised LLM-based exploratory approaches or traditional topic models on two datasets. While LLM-based methods generate more human-readable topics and show higher average win probabilities than traditional models for data exploration, they produce overly generic topics for domain-specific datasets that do not easily allow users to learn much about the documents. Adding human supervision to the LLM generation process improves data exploration by mitigating hallucination and over-genericity but requires greater human effort. In contrast, traditional. models like Latent Dirichlet Allocation (LDA) remain effective for exploration but are less user-friendly. We show that LLMs struggle to describe the haystack of large corpora without human help, particularly domain-specific data, and face scaling and hallucination limitations due to context length constraints. Dataset available at https://huggingface. co/datasets/zli12321/Bills.
2502.14752
TritonBench: Benchmarking Large Language Model Capabilities for Generating Triton Operators
cs.CL cs.LG
Triton, a high-level Python-like language designed for building efficient GPU kernels, is widely adopted in deep learning frameworks due to its portability, flexibility, and accessibility. However, programming and parallel optimization still require considerable trial and error from Triton developers. Despite advances in large language models (LLMs) for conventional code generation, these models struggle to generate accurate, performance-optimized Triton code, as they lack awareness of its specifications and the complexities of GPU programming. More critically, there is an urgent need for systematic evaluations tailored to Triton. In this work, we introduce TritonBench, the first comprehensive benchmark for Triton operator generation. TritonBench features two evaluation channels: a curated set of 184 real-world operators from GitHub and a collection of operators aligned with PyTorch interfaces. Unlike conventional code benchmarks prioritizing functional correctness, TritonBench also profiles efficiency performance on widely deployed GPUs aligned with industry applications. Our study reveals that current state-of-the-art code LLMs struggle to generate efficient Triton operators, highlighting a significant gap in high-performance code generation. TritonBench will be available at https://github.com/thunlp/TritonBench.
2502.14753
MedVAE: Efficient Automated Interpretation of Medical Images with Large-Scale Generalizable Autoencoders
eess.IV cs.AI cs.CV
Medical images are acquired at high resolutions with large fields of view in order to capture fine-grained features necessary for clinical decision-making. Consequently, training deep learning models on medical images can incur large computational costs. In this work, we address the challenge of downsizing medical images in order to improve downstream computational efficiency while preserving clinically-relevant features. We introduce MedVAE, a family of six large-scale 2D and 3D autoencoders capable of encoding medical images as downsized latent representations and decoding latent representations back to high-resolution images. We train MedVAE autoencoders using a novel two-stage training approach with 1,052,730 medical images. Across diverse tasks obtained from 20 medical image datasets, we demonstrate that (1) utilizing MedVAE latent representations in place of high-resolution images when training downstream models can lead to efficiency benefits (up to 70x improvement in throughput) while simultaneously preserving clinically-relevant features and (2) MedVAE can decode latent representations back to high-resolution images with high fidelity. Our work demonstrates that large-scale, generalizable autoencoders can help address critical efficiency challenges in the medical domain. Our code is available at https://github.com/StanfordMIMI/MedVAE.
2502.14755
Multi-Objective Causal Bayesian Optimization
stat.ML cs.LG
In decision-making problems, the outcome of an intervention often depends on the causal relationships between system components and is highly costly to evaluate. In such settings, causal Bayesian optimization (CBO) can exploit the causal relationships between the system variables and sequentially perform interventions to approach the optimum with minimal data. Extending CBO to the multi-outcome setting, we propose Multi-Objective Causal Bayesian Optimization (MO-CBO), a paradigm for identifying Pareto-optimal interventions within a known multi-target causal graph. We first derive a graphical characterization for potentially optimal sets of variables to intervene upon. Showing that any MO-CBO problem can be decomposed into several traditional multi-objective optimization tasks, we then introduce an algorithm that sequentially balances exploration across these tasks using relative hypervolume improvement. The proposed method will be validated on both synthetic and real-world causal graphs, demonstrating its superiority over traditional (non-causal) multi-objective Bayesian optimization in settings where causal information is available.
2502.14759
On the Influence of Context Size and Model Choice in Retrieval-Augmented Generation Systems
cs.CL cs.AI
Retrieval-augmented generation (RAG) has emerged as an approach to augment large language models (LLMs) by reducing their reliance on static knowledge and improving answer factuality. RAG retrieves relevant context snippets and generates an answer based on them. Despite its increasing industrial adoption, systematic exploration of RAG components is lacking, particularly regarding the ideal size of provided context, and the choice of base LLM and retrieval method. To help guide development of robust RAG systems, we evaluate various context sizes, BM25 and semantic search as retrievers, and eight base LLMs. Moving away from the usual RAG evaluation with short answers, we explore the more challenging long-form question answering in two domains, where a good answer has to utilize the entire context. Our findings indicate that final QA performance improves steadily with up to 15 snippets but stagnates or declines beyond that. Finally, we show that different general-purpose LLMs excel in the biomedical domain than the encyclopedic one, and that open-domain evidence retrieval in large corpora is challenging.
2502.14760
EquivaMap: Leveraging LLMs for Automatic Equivalence Checking of Optimization Formulations
cs.AI cs.LG math.OC
A fundamental problem in combinatorial optimization is identifying equivalent formulations, which can lead to more efficient solution strategies and deeper insights into a problem's computational complexity. The need to automatically identify equivalence between problem formulations has grown as optimization copilots--systems that generate problem formulations from natural language descriptions--have proliferated. However, existing approaches to checking formulation equivalence lack grounding, relying on simple heuristics which are insufficient for rigorous validation. Inspired by Karp reductions, in this work we introduce quasi-Karp equivalence, a formal criterion for determining when two optimization formulations are equivalent based on the existence of a mapping between their decision variables. We propose EquivaMap, a framework that leverages large language models to automatically discover such mappings, enabling scalable and reliable equivalence verification. To evaluate our approach, we construct the first open-source dataset of equivalent optimization formulations, generated by applying transformations such as adding slack variables or valid inequalities to existing formulations. Empirically, EquivaMap significantly outperforms existing methods, achieving substantial improvements in correctly identifying formulation equivalence.
2502.14762
Sculpting [CLS] Features for Pre-Trained Model-Based Class-Incremental Learning
cs.LG cs.CV
Class-incremental learning requires models to continually acquire knowledge of new classes without forgetting old ones. Although pre-trained models have demonstrated strong performance in class-incremental learning, they remain susceptible to catastrophic forgetting when learning new concepts. Excessive plasticity in the models breaks generalizability and causes forgetting, while strong stability results in insufficient adaptation to new classes. This necessitates effective adaptation with minimal modifications to preserve the general knowledge of pre-trained models. To address this challenge, we first introduce a new parameter-efficient fine-tuning module 'Learn and Calibrate', or LuCA, designed to acquire knowledge through an adapter-calibrator couple, enabling effective adaptation with well-refined feature representations. Second, for each learning session, we deploy a sparse LuCA module on top of the last token just before the classifier, which we refer to as 'Token-level Sparse Calibration and Adaptation', or TOSCA. This strategic design improves the orthogonality between the modules and significantly reduces both training and inference complexity. By leaving the generalization capabilities of the pre-trained models intact and adapting exclusively via the last token, our approach achieves a harmonious balance between stability and plasticity. Extensive experiments demonstrate TOSCA's state-of-the-art performance while introducing ~8 times fewer parameters compared to prior methods.
2502.14764
The illusion of households as entities in social networks
cs.SI physics.soc-ph
Data recording connections between people in communities and villages are collected and analyzed in various ways, most often as either networks of individuals or as networks of households. These two networks can differ in substantial ways. The methodological choice of which network to study, therefore, is an important aspect in both study design and data analysis. In this work, we consider various key differences between household and individual social network structure, and ways in which the networks cannot be used interchangeably. In addition to formalizing the choices for representing each network, we explore the consequences of how the results of social network analysis change depending on the choice between studying the individual and household network -- from determining whether networks are assortative or disassortative to the ranking of influence-maximizing nodes. As our main contribution, we draw upon related work to propose a set of systematic recommendations for determining the relevant network representation to study. Our recommendations include assessing a series of entitativity criteria and relating these criteria to theories and observations about patterns and norms in social dynamics at the household level: notably, how information spreads within households and how power structures and gender roles affect this spread. We draw upon the definition of an illusion of entitativity to identify cases wherein grouping people into households does not satisfy these criteria or adequately represent given cultural or experimental contexts. Given the widespread use of social network data for studying communities, there is broad impact in understanding which network to study and the consequences of that decision. We hope that this work gives guidance to practitioners and researchers collecting and studying social network data.
2502.14765
Step-by-Step Fact Verification System for Medical Claims with Explainable Reasoning
cs.CL cs.AI
Fact verification (FV) aims to assess the veracity of a claim based on relevant evidence. The traditional approach for automated FV includes a three-part pipeline relying on short evidence snippets and encoder-only inference models. More recent approaches leverage the multi-turn nature of LLMs to address FV as a step-by-step problem where questions inquiring additional context are generated and answered until there is enough information to make a decision. This iterative method makes the verification process rational and explainable. While these methods have been tested for encyclopedic claims, exploration on domain-specific and realistic claims is missing. In this work, we apply an iterative FV system on three medical fact-checking datasets and evaluate it with multiple settings, including different LLMs, external web search, and structured reasoning using logic predicates. We demonstrate improvements in the final performance over traditional approaches and the high potential of step-by-step FV systems for domain-specific claims.
2502.14767
Tree-of-Debate: Multi-Persona Debate Trees Elicit Critical Thinking for Scientific Comparative Analysis
cs.CL cs.AI
With the exponential growth of research facilitated by modern technology and improved accessibility, scientific discoveries have become increasingly fragmented within and across fields. This makes it challenging to assess the significance, novelty, incremental findings, and equivalent ideas between related works, particularly those from different research communities. Large language models (LLMs) have recently demonstrated strong quantitative and qualitative reasoning abilities, and multi-agent LLM debates have shown promise in handling complex reasoning tasks by exploring diverse perspectives and reasoning paths. Inspired by this, we introduce Tree-of-Debate (ToD), a framework which converts scientific papers into LLM personas that debate their respective novelties. To emphasize structured, critical reasoning rather than focusing solely on outcomes, ToD dynamically constructs a debate tree, enabling fine-grained analysis of independent novelty arguments within scholarly articles. Through experiments on scientific literature across various domains, evaluated by expert researchers, we demonstrate that ToD generates informative arguments, effectively contrasts papers, and supports researchers in their literature review.
2502.14768
Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement Learning
cs.CL cs.AI
Inspired by the success of DeepSeek-R1, we explore the potential of rule-based reinforcement learning (RL) in large reasoning models. To analyze reasoning dynamics, we use synthetic logic puzzles as training data due to their controllable complexity and straightforward answer verification. We make some key technical contributions that lead to effective and stable RL training: a system prompt that emphasizes the thinking and answering process, a stringent format reward function that penalizes outputs for taking shortcuts, and a straightforward training recipe that achieves stable convergence. Our 7B model develops advanced reasoning skills-such as reflection, verification, and summarization-that are absent from the logic corpus. Remarkably, after training on just 5K logic problems, it demonstrates generalization abilities to the challenging math benchmarks AIME and AMC.
2502.14770
Determining Layer-wise Sparsity for Large Language Models Through a Theoretical Perspective
cs.LG
In this paper, we address the challenge of determining the layer-wise sparsity rates of large language models (LLMs) through a theoretical perspective. Specifically, we identify a critical issue of ''$\textbf{reconstruction error explosion}$'' in existing LLMs sparsification methods. This refers to the cumulative effect of reconstruction errors throughout the sparsification process, where errors from earlier layers propagate and amplify in subsequent layers. As a result, the overall reconstruction error increases significantly, leading to a substantial degradation in model performance. Through theoretical analysis, we derive a simple yet effective approach to layer-wise sparsity allocation that mitigates this issue. Our method uses a monotonically increasing arithmetic progression, reducing the process of determining sparsity rates for multiple layers to the determination of a single common difference hyperparameter. Remarkably, this allows for the optimal layer-wise sparsity rates to be identified with just a few trials. Both our theoretical analysis and experimental results demonstrate that this sparsity allocation scheme is near optimal. Extensive experiments show that our method significantly improves the performance of sparse LLMs across various architectures, outperforming existing layer-wise sparsity methods. Furthermore, it enhances the performance of various compression techniques and is applicable to vision and multimodal models. Notably, our method achieves a reduction of 52.10 in perplexity for the 70$\%$ sparse LLaMA2-7B model obtained via Wanda, improves average zero-shot accuracy by 10.50$\%$, and delivers speedups of 2.63$\times$ and 2.23$\times$ on CPU and GPU, respectively.
2502.14772
Efficient Multivariate Robust Mean Estimation Under Mean-Shift Contamination
cs.DS cs.LG math.ST stat.ML stat.TH
We study the algorithmic problem of robust mean estimation of an identity covariance Gaussian in the presence of mean-shift contamination. In this contamination model, we are given a set of points in $\mathbb{R}^d$ generated i.i.d. via the following process. For a parameter $\alpha<1/2$, the $i$-th sample $x_i$ is obtained as follows: with probability $1-\alpha$, $x_i$ is drawn from $\mathcal{N}(\mu, I)$, where $\mu \in \mathbb{R}^d$ is the target mean; and with probability $\alpha$, $x_i$ is drawn from $\mathcal{N}(z_i, I)$, where $z_i$ is unknown and potentially arbitrary. Prior work characterized the information-theoretic limits of this task. Specifically, it was shown that, in contrast to Huber contamination, in the presence of mean-shift contamination consistent estimation is possible. On the other hand, all known robust estimators in the mean-shift model have running times exponential in the dimension. Here we give the first computationally efficient algorithm for high-dimensional robust mean estimation with mean-shift contamination that can tolerate a constant fraction of outliers. In particular, our algorithm has near-optimal sample complexity, runs in sample-polynomial time, and approximates the target mean to any desired accuracy. Conceptually, our result contributes to a growing body of work that studies inference with respect to natural noise models lying in between fully adversarial and random settings.
2502.14773
Sparse Activations as Conformal Predictors
cs.LG
Conformal prediction is a distribution-free framework for uncertainty quantification that replaces point predictions with sets, offering marginal coverage guarantees (i.e., ensuring that the prediction sets contain the true label with a specified probability, in expectation). In this paper, we uncover a novel connection between conformal prediction and sparse softmax-like transformations, such as sparsemax and $\gamma$-entmax (with $\gamma > 1$), which may assign nonzero probability only to a subset of labels. We introduce new non-conformity scores for classification that make the calibration process correspond to the widely used temperature scaling method. At test time, applying these sparse transformations with the calibrated temperature leads to a support set (i.e., the set of labels with nonzero probability) that automatically inherits the coverage guarantees of conformal prediction. Through experiments on computer vision and text classification benchmarks, we demonstrate that the proposed method achieves competitive results in terms of coverage, efficiency, and adaptiveness compared to standard non-conformity scores based on softmax.
2502.14776
SurveyX: Academic Survey Automation via Large Language Models
cs.CL
Large Language Models (LLMs) have demonstrated exceptional comprehension capabilities and a vast knowledge base, suggesting that LLMs can serve as efficient tools for automated survey generation. However, recent research related to automated survey generation remains constrained by some critical limitations like finite context window, lack of in-depth content discussion, and absence of systematic evaluation frameworks. Inspired by human writing processes, we propose SurveyX, an efficient and organized system for automated survey generation that decomposes the survey composing process into two phases: the Preparation and Generation phases. By innovatively introducing online reference retrieval, a pre-processing method called AttributeTree, and a re-polishing process, SurveyX significantly enhances the efficacy of survey composition. Experimental evaluation results show that SurveyX outperforms existing automated survey generation systems in content quality (0.259 improvement) and citation quality (1.76 enhancement), approaching human expert performance across multiple evaluation dimensions. Examples of surveys generated by SurveyX are available on www.surveyx.cn
2502.14777
Making Universal Policies Universal
cs.AI
The development of a generalist agent capable of solving a wide range of sequential decision-making tasks remains a significant challenge. We address this problem in a cross-agent setup where agents share the same observation space but differ in their action spaces. Our approach builds on the universal policy framework, which decouples policy learning into two stages: a diffusion-based planner that generates observation sequences and an inverse dynamics model that assigns actions to these plans. We propose a method for training the planner on a joint dataset composed of trajectories from all agents. This method offers the benefit of positive transfer by pooling data from different agents, while the primary challenge lies in adapting shared plans to each agent's unique constraints. We evaluate our approach on the BabyAI environment, covering tasks of varying complexity, and demonstrate positive transfer across agents. Additionally, we examine the planner's generalisation ability to unseen agents and compare our method to traditional imitation learning approaches. By training on a pooled dataset from multiple agents, our universal policy achieves an improvement of up to $42.20\%$ in task completion accuracy compared to a policy trained on a dataset from a single agent.
2502.14778
Harnessing PDF Data for Improving Japanese Large Multimodal Models
cs.CL cs.AI cs.CV
Large Multimodal Models (LMMs) have demonstrated strong performance in English, but their effectiveness in Japanese remains limited due to the lack of high-quality training data. Current Japanese LMMs often rely on translated English datasets, restricting their ability to capture Japan-specific cultural knowledge. To address this, we explore the potential of Japanese PDF data as a training resource, an area that remains largely underutilized. We introduce a fully automated pipeline that leverages pretrained models to extract image-text pairs from PDFs through layout analysis, OCR, and vision-language pairing, removing the need for manual annotation. Additionally, we construct instruction data from extracted image-text pairs to enrich the training data. To evaluate the effectiveness of PDF-derived data, we train Japanese LMMs and assess their performance on the Japanese LMM Benchmark. Our results demonstrate substantial improvements, with performance gains ranging from 3.9% to 13.8% on Heron-Bench. Further analysis highlights the impact of PDF-derived data on various factors, such as model size and language models, reinforcing its value as a multimodal resource for Japanese LMMs. We plan to make the source code and data publicly available upon acceptance.
2502.14779
DC-ControlNet: Decoupling Inter- and Intra-Element Conditions in Image Generation with Diffusion Models
cs.CV
In this paper, we introduce DC (Decouple)-ControlNet, a highly flexible and precisely controllable framework for multi-condition image generation. The core idea behind DC-ControlNet is to decouple control conditions, transforming global control into a hierarchical system that integrates distinct elements, contents, and layouts. This enables users to mix these individual conditions with greater flexibility, leading to more efficient and accurate image generation control. Previous ControlNet-based models rely solely on global conditions, which affect the entire image and lack the ability of element- or region-specific control. This limitation reduces flexibility and can cause condition misunderstandings in multi-conditional image generation. To address these challenges, we propose both intra-element and Inter-element Controllers in DC-ControlNet. The Intra-Element Controller handles different types of control signals within individual elements, accurately describing the content and layout characteristics of the object. For interactions between elements, we introduce the Inter-Element Controller, which accurately handles multi-element interactions and occlusion based on user-defined relationships. Extensive evaluations show that DC-ControlNet significantly outperforms existing ControlNet models and Layout-to-Image generative models in terms of control flexibility and precision in multi-condition control.
2502.14780
ReVision: A Dataset and Baseline VLM for Privacy-Preserving Task-Oriented Visual Instruction Rewriting
cs.CL cs.AI cs.CV
Efficient and privacy-preserving multimodal interaction is essential as AR, VR, and modern smartphones with powerful cameras become primary interfaces for human-computer communication. Existing powerful large vision-language models (VLMs) enabling multimodal interaction often rely on cloud-based processing, raising significant concerns about (1) visual privacy by transmitting sensitive vision data to servers, and (2) their limited real-time, on-device usability. This paper explores Visual Instruction Rewriting, a novel approach that transforms multimodal instructions into text-only commands, allowing seamless integration of lightweight on-device instruction rewriter VLMs (250M parameters) with existing conversational AI systems, enhancing vision data privacy. To achieve this, we present a dataset of over 39,000 examples across 14 domains and develop a compact VLM, pretrained on image captioning datasets and fine-tuned for instruction rewriting. Experimental results, evaluated through NLG metrics such as BLEU, METEOR, and ROUGE, along with semantic parsing analysis, demonstrate that even a quantized version of the model (<500MB storage footprint) can achieve effective instruction rewriting, thus enabling privacy-focused, multimodal AI applications.
2502.14782
A Neural Operator-Based Emulator for Regional Shallow Water Dynamics
cs.CE cs.LG physics.comp-ph physics.geo-ph
Coastal regions are particularly vulnerable to the impacts of rising sea levels and extreme weather events. Accurate real-time forecasting of hydrodynamic processes in these areas is essential for infrastructure planning and climate adaptation. In this study, we present the Multiple-Input Temporal Operator Network (MITONet), a novel autoregressive neural emulator that employs dimensionality reduction to efficiently approximate high-dimensional numerical solvers for complex, nonlinear problems that are governed by time-dependent, parameterized partial differential equations. Although MITONet is applicable to a wide range of problems, we showcase its capabilities by forecasting regional tide-driven dynamics described by the two-dimensional shallow-water equations, while incorporating initial conditions, boundary conditions, and a varying domain parameter. We demonstrate MITONet's performance in a real-world application, highlighting its ability to make accurate predictions by extrapolating both in time and parametric space.
2502.14783
Tracking and Assigning Jobs to a Markov Machine
cs.IT cs.NI cs.SY eess.SY math.IT
We consider a time-slotted communication system with a machine, a cloud server, and a sampler. Job requests from the users are queued on the server to be completed by the machine. The machine has two states, namely, a busy state and a free state. The server can assign a job to the machine in a first-in-first-served manner. If the machine is free, it completes the job request from the server; otherwise, it drops the request. Upon dropping a job request, the server is penalized. When the machine is in the free state, the machine can get into the busy state with an internal job. When the server does not assign a job request to the machine, the state of the machine evolves as a symmetric Markov chain. If the machine successfully accepts the job request from the server, the state of the machine goes to the busy state and follows a different dynamics compared to the dynamics when the machine goes to the busy state due to an internal job. The sampler samples the state of the machine and sends it to the server via an error-free channel. Thus, the server can estimate the state of the machine, upon receiving an update from the source. If the machine is in the free state but the estimated state at the server is busy, the sampler pays a cost. We incorporate the concept of the age of incorrect information to model the cost of the sampler. We aim to find an optimal sampling policy such that the cost of the sampler plus the penalty on the machine gets minimized. We formulate this problem in a Markov decision process framework and find how an optimal policy changes with several associated parameters. We show that a threshold policy is optimal for this problem. We show a necessary and sufficient condition for a threshold policy to be optimal. Finally, we find the optimal threshold without bounding the state space.
2502.14785
Real-Time Device Reach Forecasting Using HLL and MinHash Data Sketches
cs.DB cs.AI cs.LG
Predicting the right number of TVs (Device Reach) in real-time based on a user-specified targeting attributes is imperative for running multi-million dollar ADs business. The traditional approach of SQL queries to join billions of records across multiple targeting dimensions is extremely slow. As a workaround, many applications will have an offline process to crunch these numbers and present the results after many hours. In our case, the solution was an offline process taking 24 hours to onboard a customer resulting in a potential loss of business. To solve this problem, we have built a new real-time prediction system using MinHash and HyperLogLog (HLL) data sketches to compute the device reach at runtime when a user makes a request. However, existing MinHash implementations do not solve the complex problem of multilevel aggregation and intersection. This work will show how we have solved this problem, in addition, we have improved MinHash algorithm to run 4 times faster using Single Instruction Multiple Data (SIMD) vectorized operations for high speed and accuracy with constant space to process billions of records. Finally, by experiments, we prove that the results are as accurate as traditional offline prediction system with an acceptable error rate of 5%.
2502.14786
SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features
cs.CV cs.AI
We introduce SigLIP 2, a family of new multilingual vision-language encoders that build on the success of the original SigLIP. In this second iteration, we extend the original image-text training objective with several prior, independently developed techniques into a unified recipe -- this includes captioning-based pretraining, self-supervised losses (self-distillation, masked prediction) and online data curation. With these changes, SigLIP 2 models outperform their SigLIP counterparts at all model scales in core capabilities, including zero-shot classification, image-text retrieval, and transfer performance when extracting visual representations for Vision-Language Models (VLMs). Furthermore, the new training recipe leads to significant improvements on localization and dense prediction tasks. We also train variants which support multiple resolutions and preserve the input's native aspect ratio. Finally, we train on a more diverse data-mixture that includes de-biasing techniques, leading to much better multilingual understanding and improved fairness. To allow users to trade off inference cost with performance, we release model checkpoints at four sizes: ViT-B (86M), L (303M), So400m (400M), and g (1B).
2502.14788
Ray-Tracing for Conditionally Activated Neural Networks
cs.LG cs.AI
In this paper, we introduce a novel architecture for conditionally activated neural networks combining a hierarchical construction of multiple Mixture of Experts (MoEs) layers with a sampling mechanism that progressively converges to an optimized configuration of expert activation. This methodology enables the dynamic unfolding of the network's architecture, facilitating efficient path-specific training. Experimental results demonstrate that this approach achieves competitive accuracy compared to conventional baselines while significantly reducing the parameter count required for inference. Notably, this parameter reduction correlates with the complexity of the input patterns, a property naturally emerging from the network's operational dynamics without necessitating explicit auxiliary penalty functions.
2502.14789
Structurally Disentangled Feature Fields Distillation for 3D Understanding and Editing
cs.CV
Recent work has demonstrated the ability to leverage or distill pre-trained 2D features obtained using large pre-trained 2D models into 3D features, enabling impressive 3D editing and understanding capabilities using only 2D supervision. Although impressive, models assume that 3D features are captured using a single feature field and often make a simplifying assumption that features are view-independent. In this work, we propose instead to capture 3D features using multiple disentangled feature fields that capture different structural components of 3D features involving view-dependent and view-independent components, which can be learned from 2D feature supervision only. Subsequently, each element can be controlled in isolation, enabling semantic and structural understanding and editing capabilities. For instance, using a user click, one can segment 3D features corresponding to a given object and then segment, edit, or remove their view-dependent (reflective) properties. We evaluate our approach on the task of 3D segmentation and demonstrate a set of novel understanding and editing tasks.
2502.14790
An Adversarial Analysis of Thompson Sampling for Full-information Online Learning: from Finite to Infinite Action Spaces
cs.LG cs.GT math.ST stat.ML stat.TH
We develop an analysis of Thompson sampling for online learning under full feedback - also known as prediction with expert advice - where the learner's prior is defined over the space of an adversary's future actions, rather than the space of experts. We show regret decomposes into regret the learner expected a priori, plus a prior-robustness-type term we call excess regret. In the classical finite-expert setting, this recovers optimal rates. As an initial step towards practical online learning in settings with a potentially-uncountably-infinite number of experts, we show that Thompson sampling with a certain Gaussian process prior widely-used in the Bayesian optimization literature has a $\mathcal{O}(\beta\sqrt{T\log(1+\lambda)})$ rate against a $\beta$-bounded $\lambda$-Lipschitz~adversary.
2502.14791
Rapid Word Learning Through Meta In-Context Learning
cs.CL cs.AI cs.LG
Humans can quickly learn a new word from a few illustrative examples, and then systematically and flexibly use it in novel contexts. Yet the abilities of current language models for few-shot word learning, and methods for improving these abilities, are underexplored. In this study, we introduce a novel method, Meta-training for IN-context learNing Of Words (Minnow). This method trains language models to generate new examples of a word's usage given a few in-context examples, using a special placeholder token to represent the new word. This training is repeated on many new words to develop a general word-learning ability. We find that training models from scratch with Minnow on human-scale child-directed language enables strong few-shot word learning, comparable to a large language model (LLM) pre-trained on orders of magnitude more data. Furthermore, through discriminative and generative evaluations, we demonstrate that finetuning pre-trained LLMs with Minnow improves their ability to discriminate between new words, identify syntactic categories of new words, and generate reasonable new usages and definitions for new words, based on one or a few in-context examples. These findings highlight the data efficiency of Minnow and its potential to improve language model performance in word learning tasks.
2502.14792
RendBEV: Semantic Novel View Synthesis for Self-Supervised Bird's Eye View Segmentation
cs.CV
Bird's Eye View (BEV) semantic maps have recently garnered a lot of attention as a useful representation of the environment to tackle assisted and autonomous driving tasks. However, most of the existing work focuses on the fully supervised setting, training networks on large annotated datasets. In this work, we present RendBEV, a new method for the self-supervised training of BEV semantic segmentation networks, leveraging differentiable volumetric rendering to receive supervision from semantic perspective views computed by a 2D semantic segmentation model. Our method enables zero-shot BEV semantic segmentation, and already delivers competitive results in this challenging setting. When used as pretraining to then fine-tune on labeled BEV ground-truth, our method significantly boosts performance in low-annotation regimes, and sets a new state of the art when fine-tuning on all available labels.
2502.14795
Humanoid-VLA: Towards Universal Humanoid Control with Visual Integration
cs.RO cs.CV
This paper addresses the limitations of current humanoid robot control frameworks, which primarily rely on reactive mechanisms and lack autonomous interaction capabilities due to data scarcity. We propose Humanoid-VLA, a novel framework that integrates language understanding, egocentric scene perception, and motion control, enabling universal humanoid control. Humanoid-VLA begins with language-motion pre-alignment using non-egocentric human motion datasets paired with textual descriptions, allowing the model to learn universal motion patterns and action semantics. We then incorporate egocentric visual context through a parameter efficient video-conditioned fine-tuning, enabling context-aware motion generation. Furthermore, we introduce a self-supervised data augmentation strategy that automatically generates pseudoannotations directly derived from motion data. This process converts raw motion sequences into informative question-answer pairs, facilitating the effective use of large-scale unlabeled video data. Built upon whole-body control architectures, extensive experiments show that Humanoid-VLA achieves object interaction and environment exploration tasks with enhanced contextual awareness, demonstrating a more human-like capacity for adaptive and intelligent engagement.
2502.14796
A Multi-Agent Perspective on Modern Information Retrieval
cs.IR
The rise of large language models (LLMs) has introduced a new era in information retrieval (IR), where queries and documents that were once assumed to be generated exclusively by humans can now also be created by automated agents. These agents can formulate queries, generate documents, and perform ranking. This shift challenges some long-standing IR paradigms and calls for a reassessment of both theoretical frameworks and practical methodologies. We advocate for a multi-agent perspective to better capture the complex interactions between query agents, document agents, and ranker agents. Through empirical exploration of various multi-agent retrieval settings, we reveal the significant impact of these interactions on system performance. Our findings underscore the need to revisit classical IR paradigms and develop new frameworks for more effective modeling and evaluation of modern retrieval systems.
2502.14799
A Survey on Text-Driven 360-Degree Panorama Generation
cs.CV cs.AI
The advent of text-driven 360-degree panorama generation, enabling the synthesis of 360-degree panoramic images directly from textual descriptions, marks a transformative advancement in immersive visual content creation. This innovation significantly simplifies the traditionally complex process of producing such content. Recent progress in text-to-image diffusion models has accelerated the rapid development in this emerging field. This survey presents a comprehensive review of text-driven 360-degree panorama generation, offering an in-depth analysis of state-of-the-art algorithms and their expanding applications in 360-degree 3D scene generation. Furthermore, we critically examine current limitations and propose promising directions for future research. A curated project page with relevant resources and research papers is available at https://littlewhitesea.github.io/Text-Driven-Pano-Gen/.
2502.14801
AVD2: Accident Video Diffusion for Accident Video Description
cs.CV
Traffic accidents present complex challenges for autonomous driving, often featuring unpredictable scenarios that hinder accurate system interpretation and responses.Nonetheless, prevailing methodologies fall short in elucidating the causes of accidents and proposing preventive measures due to the paucity of training data specific to accident scenarios.In this work, we introduce AVD2 (Accident Video Diffusion for Accident Video Description), a novel framework that enhances accident scene understanding by generating accident videos that aligned with detailed natural language descriptions and reasoning, resulting in the contributed EMM-AU (Enhanced Multi-Modal Accident Video Understanding) dataset. Empirical results reveal that the integration of the EMM-AU dataset establishes state-of-the-art performance across both automated metrics and human evaluations, markedly advancing the domains of accident analysis and prevention. Project resources are available at https://an-answer-tree.github.io
2502.14802
From RAG to Memory: Non-Parametric Continual Learning for Large Language Models
cs.CL cs.AI
Our ability to continuously acquire, organize, and leverage knowledge is a key feature of human intelligence that AI systems must approximate to unlock their full potential. Given the challenges in continual learning with large language models (LLMs), retrieval-augmented generation (RAG) has become the dominant way to introduce new information. However, its reliance on vector retrieval hinders its ability to mimic the dynamic and interconnected nature of human long-term memory. Recent RAG approaches augment vector embeddings with various structures like knowledge graphs to address some of these gaps, namely sense-making and associativity. However, their performance on more basic factual memory tasks drops considerably below standard RAG. We address this unintended deterioration and propose HippoRAG 2, a framework that outperforms standard RAG comprehensively on factual, sense-making, and associative memory tasks. HippoRAG 2 builds upon the Personalized PageRank algorithm used in HippoRAG and enhances it with deeper passage integration and more effective online use of an LLM. This combination pushes this RAG system closer to the effectiveness of human long-term memory, achieving a 7% improvement in associative memory tasks over the state-of-the-art embedding model while also exhibiting superior factual knowledge and sense-making memory capabilities. This work paves the way for non-parametric continual learning for LLMs. Our code and data will be released at https://github.com/OSU-NLP-Group/HippoRAG.
2502.14803
Planning, scheduling, and execution on the Moon: the CADRE technology demonstration mission
cs.RO cs.SY eess.SY
NASA's Cooperative Autonomous Distributed Robotic Exploration (CADRE) mission, slated for flight to the Moon's Reiner Gamma region in 2025/2026, is designed to demonstrate multi-agent autonomous exploration of the Lunar surface and sub-surface. A team of three robots and a base station will autonomously explore a region near the lander, collecting the data required for 3D reconstruction of the surface with no human input; and then autonomously perform distributed sensing with multi-static ground penetrating radars (GPR), driving in formation while performing coordinated radar soundings to create a map of the subsurface. At the core of CADRE's software architecture is a novel autonomous, distributed planning, scheduling, and execution (PS&E) system. The system coordinates the robots' activities, planning and executing tasks that require multiple robots' participation while ensuring that each individual robot's thermal and power resources stay within prescribed bounds, and respecting ground-prescribed sleep-wake cycles. The system uses a centralized-planning, distributed-execution paradigm, and a leader election mechanism ensures robustness to failures of individual agents. In this paper, we describe the architecture of CADRE's PS&E system; discuss its design rationale; and report on verification and validation (V&V) testing of the system on CADRE's hardware in preparation for deployment on the Moon.
2502.14807
FetalCLIP: A Visual-Language Foundation Model for Fetal Ultrasound Image Analysis
eess.IV cs.AI cs.CV
Foundation models are becoming increasingly effective in the medical domain, offering pre-trained models on large datasets that can be readily adapted for downstream tasks. Despite progress, fetal ultrasound images remain a challenging domain for foundation models due to their inherent complexity, often requiring substantial additional training and facing limitations due to the scarcity of paired multimodal data. To overcome these challenges, here we introduce FetalCLIP, a vision-language foundation model capable of generating universal representation of fetal ultrasound images. FetalCLIP was pre-trained using a multimodal learning approach on a diverse dataset of 210,035 fetal ultrasound images paired with text. This represents the largest paired dataset of its kind used for foundation model development to date. This unique training approach allows FetalCLIP to effectively learn the intricate anatomical features present in fetal ultrasound images, resulting in robust representations that can be used for a variety of downstream applications. In extensive benchmarking across a range of key fetal ultrasound applications, including classification, gestational age estimation, congenital heart defect (CHD) detection, and fetal structure segmentation, FetalCLIP outperformed all baselines while demonstrating remarkable generalizability and strong performance even with limited labeled data. We plan to release the FetalCLIP model publicly for the benefit of the broader scientific community.
2502.14809
PREM: Privately Answering Statistical Queries with Relative Error
cs.LG
We introduce $\mathsf{PREM}$ (Private Relative Error Multiplicative weight update), a new framework for generating synthetic data that achieves a relative error guarantee for statistical queries under $(\varepsilon, \delta)$ differential privacy (DP). Namely, for a domain ${\cal X}$, a family ${\cal F}$ of queries $f : {\cal X} \to \{0, 1\}$, and $\zeta > 0$, our framework yields a mechanism that on input dataset $D \in {\cal X}^n$ outputs a synthetic dataset $\widehat{D} \in {\cal X}^n$ such that all statistical queries in ${\cal F}$ on $D$, namely $\sum_{x \in D} f(x)$ for $f \in {\cal F}$, are within a $1 \pm \zeta$ multiplicative factor of the corresponding value on $\widehat{D}$ up to an additive error that is polynomial in $\log |{\cal F}|$, $\log |{\cal X}|$, $\log n$, $\log(1/\delta)$, $1/\varepsilon$, and $1/\zeta$. In contrast, any $(\varepsilon, \delta)$-DP mechanism is known to require worst-case additive error that is polynomial in at least one of $n, |{\cal F}|$, or $|{\cal X}|$. We complement our algorithm with nearly matching lower bounds.
2502.14814
VB-Com: Learning Vision-Blind Composite Humanoid Locomotion Against Deficient Perception
cs.RO
The performance of legged locomotion is closely tied to the accuracy and comprehensiveness of state observations. Blind policies, which rely solely on proprioception, are considered highly robust due to the reliability of proprioceptive observations. However, these policies significantly limit locomotion speed and often require collisions with the terrain to adapt. In contrast, Vision policies allows the robot to plan motions in advance and respond proactively to unstructured terrains with an online perception module. However, perception is often compromised by noisy real-world environments, potential sensor failures, and the limitations of current simulations in presenting dynamic or deformable terrains. Humanoid robots, with high degrees of freedom and inherently unstable morphology, are particularly susceptible to misguidance from deficient perception, which can result in falls or termination on challenging dynamic terrains. To leverage the advantages of both vision and blind policies, we propose VB-Com, a composite framework that enables humanoid robots to determine when to rely on the vision policy and when to switch to the blind policy under perceptual deficiency. We demonstrate that VB-Com effectively enables humanoid robots to traverse challenging terrains and obstacles despite perception deficiencies caused by dynamic terrains or perceptual noise.
2502.14815
Optimizing Model Selection for Compound AI Systems
cs.AI cs.CL cs.LG cs.MA
Compound AI systems that combine multiple LLM calls, such as self-refine and multi-agent-debate, achieve strong performance on many AI tasks. We address a core question in optimizing compound systems: for each LLM call or module in the system, how should one decide which LLM to use? We show that these LLM choices have a large effect on quality, but the search space is exponential. We propose LLMSelector, an efficient framework for model selection in compound systems, which leverages two key empirical insights: (i) end-to-end performance is often monotonic in how well each module performs, with all other modules held fixed, and (ii) per-module performance can be estimated accurately by an LLM. Building upon these insights, LLMSelector iteratively selects one module and allocates to it the model with the highest module-wise performance, as estimated by an LLM, until no further gain is possible. LLMSelector is applicable to any compound system with a bounded number of modules, and its number of API calls scales linearly with the number of modules, achieving high-quality model allocation both empirically and theoretically. Experiments with popular compound systems such as multi-agent debate and self-refine using LLMs such as GPT-4o, Claude 3.5 Sonnet and Gemini 1.5 show that LLMSelector confers 5%-70% accuracy gains compared to using the same LLM for all modules.
2502.14816
Dynamic Low-Rank Sparse Adaptation for Large Language Models
cs.LG
Despite the efficacy of network sparsity in alleviating the deployment strain of Large Language Models (LLMs), it endures significant performance degradation. Applying Low-Rank Adaptation (LoRA) to fine-tune the sparse LLMs offers an intuitive approach to counter this predicament, while it holds shortcomings include: 1) The inability to integrate LoRA weights into sparse LLMs post-training, and 2) Insufficient performance recovery at high sparsity ratios. In this paper, we introduce dynamic Low-rank Sparse Adaptation (LoSA), a novel method that seamlessly integrates low-rank adaptation into LLM sparsity within a unified framework, thereby enhancing the performance of sparse LLMs without increasing the inference latency. In particular, LoSA dynamically sparsifies the LoRA outcomes based on the corresponding sparse weights during fine-tuning, thus guaranteeing that the LoRA module can be integrated into the sparse LLMs post-training. Besides, LoSA leverages Representation Mutual Information (RMI) as an indicator to determine the importance of layers, thereby efficiently determining the layer-wise sparsity rates during fine-tuning. Predicated on this, LoSA adjusts the rank of the LoRA module based on the variability in layer-wise reconstruction errors, allocating an appropriate fine-tuning for each layer to reduce the output discrepancies between dense and sparse LLMs. Extensive experiments tell that LoSA can efficiently boost the efficacy of sparse LLMs within a few hours, without introducing any additional inferential burden. For example, LoSA reduced the perplexity of sparse LLaMA-2-7B by 68.73 and increased zero-shot accuracy by 16.32$\%$, achieving a 2.60$\times$ speedup on CPU and 2.23$\times$ speedup on GPU, requiring only 45 minutes of fine-tuning on a single NVIDIA A100 80GB GPU. Code is available at https://github.com/wzhuang-xmu/LoSA.
2502.14819
Learning from Reward-Free Offline Data: A Case for Planning with Latent Dynamics Models
cs.LG
A long-standing goal in AI is to build agents that can solve a variety of tasks across different environments, including previously unseen ones. Two dominant approaches tackle this challenge: (i) reinforcement learning (RL), which learns policies through trial and error, and (ii) optimal control, which plans actions using a learned or known dynamics model. However, their relative strengths and weaknesses remain underexplored in the setting where agents must learn from offline trajectories without reward annotations. In this work, we systematically analyze the performance of different RL and control-based methods under datasets of varying quality. On the RL side, we consider goal-conditioned and zero-shot approaches. On the control side, we train a latent dynamics model using the Joint Embedding Predictive Architecture (JEPA) and use it for planning. We study how dataset properties-such as data diversity, trajectory quality, and environment variability-affect the performance of these approaches. Our results show that model-free RL excels when abundant, high-quality data is available, while model-based planning excels in generalization to novel environment layouts, trajectory stitching, and data-efficiency. Notably, planning with a latent dynamics model emerges as a promising approach for zero-shot generalization from suboptimal data.
2502.14820
eC-Tab2Text: Aspect-Based Text Generation from e-Commerce Product Tables
cs.CL cs.AI cs.DB cs.HC
Large Language Models (LLMs) have demonstrated exceptional versatility across diverse domains, yet their application in e-commerce remains underexplored due to a lack of domain-specific datasets. To address this gap, we introduce eC-Tab2Text, a novel dataset designed to capture the intricacies of e-commerce, including detailed product attributes and user-specific queries. Leveraging eC-Tab2Text, we focus on text generation from product tables, enabling LLMs to produce high-quality, attribute-specific product reviews from structured tabular data. Fine-tuned models were rigorously evaluated using standard Table2Text metrics, alongside correctness, faithfulness, and fluency assessments. Our results demonstrate substantial improvements in generating contextually accurate reviews, highlighting the transformative potential of tailored datasets and fine-tuning methodologies in optimizing e-commerce workflows. This work highlights the potential of LLMs in e-commerce workflows and the essential role of domain-specific datasets in tailoring them to industry-specific challenges.
2502.14821
Meshless Shape Optimization using Neural Networks and Partial Differential Equations on Graphs
math.NA cs.LG cs.NA math.OC
Shape optimization involves the minimization of a cost function defined over a set of shapes, often governed by a partial differential equation (PDE). In the absence of closed-form solutions, one relies on numerical methods to approximate the solution. The level set method -- when coupled with the finite element method -- is one of the most versatile numerical shape optimization approaches but still suffers from the limitations of most mesh-based methods. In this work, we present a fully meshless level set framework that leverages neural networks to parameterize the level set function and employs the graph Laplacian to approximate the underlying PDE. Our approach enables precise computations of geometric quantities such as surface normals and curvature, and allows tackling optimization problems within the class of convex shapes.
2502.14822
A Survey of Model Architectures in Information Retrieval
cs.IR
This survey examines the evolution of model architectures in information retrieval (IR), focusing on two key aspects: backbone models for feature extraction and end-to-end system architectures for relevance estimation. The review intentionally separates architectural considerations from training methodologies to provide a focused analysis of structural innovations in IR systems.We trace the development from traditional term-based methods to modern neural approaches, particularly highlighting the impact of transformer-based models and subsequent large language models (LLMs). We conclude by discussing emerging challenges and future directions, including architectural optimizations for performance and scalability, handling of multimodal, multilingual data, and adaptation to novel application domains beyond traditional search paradigms.
2502.14827
Exploring Advanced Techniques for Visual Question Answering: A Comprehensive Comparison
cs.CV cs.AI cs.ET cs.LG
Visual Question Answering (VQA) has emerged as a pivotal task in the intersection of computer vision and natural language processing, requiring models to understand and reason about visual content in response to natural language questions. Analyzing VQA datasets is essential for developing robust models that can handle the complexities of multimodal reasoning. Several approaches have been developed to examine these datasets, each offering distinct perspectives on question diversity, answer distribution, and visual-textual correlations. Despite significant progress, existing VQA models face challenges related to dataset bias, limited model complexity, commonsense reasoning gaps, rigid evaluation methods, and generalization to real world scenarios. This paper presents a comprehensive comparative study of five advanced VQA models: ABC-CNN, KICNLE, Masked Vision and Language Modeling, BLIP-2, and OFA, each employing distinct methodologies to address these challenges.
2502.14828
Fundamental Limitations in Defending LLM Finetuning APIs
cs.LG cs.CR
LLM developers have imposed technical interventions to prevent fine-tuning misuse attacks, attacks where adversaries evade safeguards by fine-tuning the model using a public API. Previous work has established several successful attacks against specific fine-tuning API defences. In this work, we show that defences of fine-tuning APIs that seek to detect individual harmful training or inference samples ('pointwise' detection) are fundamentally limited in their ability to prevent fine-tuning attacks. We construct 'pointwise-undetectable' attacks that repurpose entropy in benign model outputs (e.g. semantic or syntactic variations) to covertly transmit dangerous knowledge. Our attacks are composed solely of unsuspicious benign samples that can be collected from the model before fine-tuning, meaning training and inference samples are all individually benign and low-perplexity. We test our attacks against the OpenAI fine-tuning API, finding they succeed in eliciting answers to harmful multiple-choice questions, and that they evade an enhanced monitoring system we design that successfully detects other fine-tuning attacks. We encourage the community to develop defences that tackle the fundamental limitations we uncover in pointwise fine-tuning API defences.