id
string
title
string
abstract
string
text
string
label
string
primary_category
string
published
string
2512.17688v2
Convergence Guarantees for Federated SARSA with Local Training and Heterogeneous Agents
We present a novel theoretical analysis of Federated SARSA (FedSARSA) with linear function approximation and local training. We establish convergence guarantees for FedSARSA in the presence of heterogeneity, both in local transitions and rewards, providing the first sample and communication complexity bounds in this se...
Convergence Guarantees for Federated SARSA with Local Training and Heterogeneous Agents. We present a novel theoretical analysis of Federated SARSA (FedSARSA) with linear function approximation and local training. We establish convergence guarantees for FedSARSA in the presence of heterogeneity, both in local transitio...
lg_ml
cs.LG
2025-12-19T15:23:44+00:00
2511.03797v1
Learning Paths for Dynamic Measure Transport: A Control Perspective
We bring a control perspective to the problem of identifying paths of measures for sampling via dynamic measure transport (DMT). We highlight the fact that commonly used paths may be poor choices for DMT and connect existing methods for learning alternate paths to mean-field games. Based on these connections we pose a ...
Learning Paths for Dynamic Measure Transport: A Control Perspective. We bring a control perspective to the problem of identifying paths of measures for sampling via dynamic measure transport (DMT). We highlight the fact that commonly used paths may be poor choices for DMT and connect existing methods for learning alter...
stat_ml
stat.ML
2025-11-05T19:02:23+00:00
2512.05371v1
ChipMind: Retrieval-Augmented Reasoning for Long-Context Circuit Design Specifications
While Large Language Models (LLMs) demonstrate immense potential for automating integrated circuit (IC) development, their practical deployment is fundamentally limited by restricted context windows. Existing context-extension methods struggle to achieve effective semantic modeling and thorough multi-hop reasoning over...
ChipMind: Retrieval-Augmented Reasoning for Long-Context Circuit Design Specifications. While Large Language Models (LLMs) demonstrate immense potential for automating integrated circuit (IC) development, their practical deployment is fundamentally limited by restricted context windows. Existing context-extension metho...
ai_general
cs.AI
2025-12-05T02:09:49+00:00
2512.21562v1
Exploration of Reproducible Generated Image Detection
While the technology for detecting AI-Generated Content (AIGC) images has advanced rapidly, the field still faces two core issues: poor reproducibility and insufficient gen eralizability, which hinder the practical application of such technologies. This study addresses these challenges by re viewing 7 key papers on AIG...
Exploration of Reproducible Generated Image Detection. While the technology for detecting AI-Generated Content (AIGC) images has advanced rapidly, the field still faces two core issues: poor reproducibility and insufficient gen eralizability, which hinder the practical application of such technologies. This study addre...
cv_vision
cs.CV
2025-12-25T08:16:41+00:00
2512.25023v1
ResponseRank: Data-Efficient Reward Modeling through Preference Strength Learning
Binary choices, as often used for reinforcement learning from human feedback (RLHF), convey only the direction of a preference. A person may choose apples over oranges and bananas over grapes, but which preference is stronger? Strength is crucial for decision-making under uncertainty and generalization of preference mo...
ResponseRank: Data-Efficient Reward Modeling through Preference Strength Learning. Binary choices, as often used for reinforcement learning from human feedback (RLHF), convey only the direction of a preference. A person may choose apples over oranges and bananas over grapes, but which preference is stronger? Strength i...
lg_ml
cs.LG
2025-12-31T18:21:52+00:00
2512.17278v1
WDFFU-Mamba: A Wavelet-guided Dual-attention Feature Fusion Mamba for Breast Tumor Segmentation in Ultrasound Images
Breast ultrasound (BUS) image segmentation plays a vital role in assisting clinical diagnosis and early tumor screening. However, challenges such as speckle noise, imaging artifacts, irregular lesion morphology, and blurred boundaries severely hinder accurate segmentation. To address these challenges, this work aims to...
WDFFU-Mamba: A Wavelet-guided Dual-attention Feature Fusion Mamba for Breast Tumor Segmentation in Ultrasound Images. Breast ultrasound (BUS) image segmentation plays a vital role in assisting clinical diagnosis and early tumor screening. However, challenges such as speckle noise, imaging artifacts, irregular lesion mo...
cv_vision
cs.CV
2025-12-19T06:50:03+00:00
2512.10150v1
Unforgotten Safety: Preserving Safety Alignment of Large Language Models with Continual Learning
The safety alignment of large language models (LLMs) is becoming increasingly important with their democratization. In this paper, we study the safety degradation that comes with adapting LLMs to new tasks. We attribute this safety compromise to catastrophic forgetting and frame the problem of preserving safety when fi...
Unforgotten Safety: Preserving Safety Alignment of Large Language Models with Continual Learning. The safety alignment of large language models (LLMs) is becoming increasingly important with their democratization. In this paper, we study the safety degradation that comes with adapting LLMs to new tasks. We attribute th...
cl_nlp
cs.CL
2025-12-10T23:16:47+00:00
2512.16824v1
Tiny Recursive Control: Iterative Reasoning for Efficient Optimal Control
Neural network controllers increasingly demand millions of parameters, and language model approaches push into the billions. For embedded aerospace systems with strict power and latency constraints, this scaling is prohibitive. We present Tiny Recursive Control (TRC), a neural architecture based on a counterintuitive p...
Tiny Recursive Control: Iterative Reasoning for Efficient Optimal Control. Neural network controllers increasingly demand millions of parameters, and language model approaches push into the billions. For embedded aerospace systems with strict power and latency constraints, this scaling is prohibitive. We present Tiny R...
lg_ml
cs.LG
2025-12-18T18:05:05+00:00
2601.04208v1
LLMs for Explainable Business Decision-Making: A Reinforcement Learning Fine-Tuning Approach
Artificial Intelligence (AI) models increasingly drive high-stakes consumer interactions, yet their decision logic often remains opaque. Prevailing explainable AI techniques rely on post hoc numerical feature attributions, which fail to provide coherent narratives behind model decisions. Large language models (LLMs) pr...
LLMs for Explainable Business Decision-Making: A Reinforcement Learning Fine-Tuning Approach. Artificial Intelligence (AI) models increasingly drive high-stakes consumer interactions, yet their decision logic often remains opaque. Prevailing explainable AI techniques rely on post hoc numerical feature attributions, whi...
cl_nlp
cs.CL
2025-12-10T04:16:31+00:00
2512.19720v1
Per-Axis Weight Deltas for Frequent Model Updates
Serving many task-specialized LLM variants is often limited by the large size of fine-tuned checkpoints and the resulting cold-start latency. Since fine-tuned weights differ from their base model by relatively small structured residuals, a natural approach is to represent them as compressed deltas. We propose a simple ...
Per-Axis Weight Deltas for Frequent Model Updates. Serving many task-specialized LLM variants is often limited by the large size of fine-tuned checkpoints and the resulting cold-start latency. Since fine-tuned weights differ from their base model by relatively small structured residuals, a natural approach is to repres...
lg_ml
cs.LG
2025-12-16T16:46:28+00:00
2512.15606v1
A Teacher-Student Perspective on the Dynamics of Learning Near the Optimal Point
Near an optimal learning point of a neural network, the learning performance of gradient descent dynamics is dictated by the Hessian matrix of the loss function with respect to the network parameters. We characterize the Hessian eigenspectrum for some classes of teacher-student problems, when the teacher and student ne...
A Teacher-Student Perspective on the Dynamics of Learning Near the Optimal Point. Near an optimal learning point of a neural network, the learning performance of gradient descent dynamics is dictated by the Hessian matrix of the loss function with respect to the network parameters. We characterize the Hessian eigenspec...
stat_ml
stat.ML
2025-12-17T17:17:12+00:00
2510.10511v2
Enhancing Long-Term Welfare in Recommender Systems: An Information Revelation Approach
Improving the long-term user welfare (e.g., sustained user engagement) has become a central objective of recommender systems (RS). In real-world platforms, the creation behaviors of content creators plays a crucial role in shaping long-term welfare beyond short-term recommendation accuracy, making the effective steerin...
Enhancing Long-Term Welfare in Recommender Systems: An Information Revelation Approach. Improving the long-term user welfare (e.g., sustained user engagement) has become a central objective of recommender systems (RS). In real-world platforms, the creation behaviors of content creators plays a crucial role in shaping l...
ir_retrieval
cs.IR
2025-10-12T09:18:15+00:00
2512.04392v1
Informative missingness and its implications in semi-supervised learning
Semi-supervised learning (SSL) constructs classifiers using both labelled and unlabelled data. It leverages information from labelled samples, whose acquisition is often costly or labour-intensive, together with unlabelled data to enhance prediction performance. This defines an incomplete-data problem, which statistica...
Informative missingness and its implications in semi-supervised learning. Semi-supervised learning (SSL) constructs classifiers using both labelled and unlabelled data. It leverages information from labelled samples, whose acquisition is often costly or labour-intensive, together with unlabelled data to enhance predict...
stat_ml
stat.ML
2025-12-04T02:26:56+00:00
2512.22522v3
Towards Reliable Evaluation of Adversarial Robustness for Spiking Neural Networks
Spiking Neural Networks (SNNs) utilize spike-based activations to mimic the brain's energy-efficient information processing. However, the binary and discontinuous nature of spike activations causes vanishing gradients, making adversarial robustness evaluation via gradient descent unreliable. While improved surrogate gr...
Towards Reliable Evaluation of Adversarial Robustness for Spiking Neural Networks. Spiking Neural Networks (SNNs) utilize spike-based activations to mimic the brain's energy-efficient information processing. However, the binary and discontinuous nature of spike activations causes vanishing gradients, making adversarial...
lg_ml
cs.LG
2025-12-27T08:43:06+00:00
2512.20026v1
MAPI-GNN: Multi-Activation Plane Interaction Graph Neural Network for Multimodal Medical Diagnosis
Graph neural networks are increasingly applied to multimodal medical diagnosis for their inherent relational modeling capabilities. However, their efficacy is often compromised by the prevailing reliance on a single, static graph built from indiscriminate features, hindering the ability to model patient-specific pathol...
MAPI-GNN: Multi-Activation Plane Interaction Graph Neural Network for Multimodal Medical Diagnosis. Graph neural networks are increasingly applied to multimodal medical diagnosis for their inherent relational modeling capabilities. However, their efficacy is often compromised by the prevailing reliance on a single, sta...
cv_vision
cs.CV
2025-12-23T03:38:57+00:00
2512.21211v1
Causal-driven attribution (CDA): Estimating channel influence without user-level data
Attribution modelling lies at the heart of marketing effectiveness, yet most existing approaches depend on user-level path data, which are increasingly inaccessible due to privacy regulations and platform restrictions. This paper introduces a Causal-Driven Attribution (CDA) framework that infers channel influence using...
Causal-driven attribution (CDA): Estimating channel influence without user-level data. Attribution modelling lies at the heart of marketing effectiveness, yet most existing approaches depend on user-level path data, which are increasingly inaccessible due to privacy regulations and platform restrictions. This paper int...
stat_ml
stat.ML
2025-12-24T14:51:12+00:00
2512.08344v1
Enhancing Explainability of Graph Neural Networks Through Conceptual and Structural Analyses and Their Extensions
Graph Neural Networks (GNNs) have become a powerful tool for modeling and analyzing data with graph structures. The wide adoption in numerous applications underscores the value of these models. However, the complexity of these methods often impedes understanding their decision-making processes. Current Explainable AI (...
Enhancing Explainability of Graph Neural Networks Through Conceptual and Structural Analyses and Their Extensions. Graph Neural Networks (GNNs) have become a powerful tool for modeling and analyzing data with graph structures. The wide adoption in numerous applications underscores the value of these models. However, th...
ai_general
cs.AI
2025-12-09T08:13:31+00:00
2512.20885v2
From GNNs to Symbolic Surrogates via Kolmogorov-Arnold Networks for Delay Prediction
Accurate prediction of flow delay is essential for optimizing and managing modern communication networks. We investigate three levels of modeling for this task. First, we implement a heterogeneous GNN with attention-based message passing, establishing a strong neural baseline. Second, we propose FlowKANet in which Kolm...
From GNNs to Symbolic Surrogates via Kolmogorov-Arnold Networks for Delay Prediction. Accurate prediction of flow delay is essential for optimizing and managing modern communication networks. We investigate three levels of modeling for this task. First, we implement a heterogeneous GNN with attention-based message pass...
lg_ml
cs.LG
2025-12-24T02:05:46+00:00
2512.23255v1
Contour Information Aware 2D Gaussian Splatting for Image Representation
Image representation is a fundamental task in computer vision. Recently, Gaussian Splatting has emerged as an efficient representation framework, and its extension to 2D image representation enables lightweight, yet expressive modeling of visual content. While recent 2D Gaussian Splatting (2DGS) approaches provide comp...
Contour Information Aware 2D Gaussian Splatting for Image Representation. Image representation is a fundamental task in computer vision. Recently, Gaussian Splatting has emerged as an efficient representation framework, and its extension to 2D image representation enables lightweight, yet expressive modeling of visual ...
cv_vision
cs.CV
2025-12-29T07:24:36+00:00
2509.15822v2
Phase Transition for Stochastic Block Model with more than $\sqrt{n}$ Communities
Predictions from statistical physics postulate that recovery of the communities in Stochastic Block Model (SBM) is possible in polynomial time above, and only above, the Kesten-Stigum (KS) threshold. This conjecture has given rise to a rich literature, proving that non-trivial community recovery is indeed possible in S...
Phase Transition for Stochastic Block Model with more than $\sqrt{n}$ Communities. Predictions from statistical physics postulate that recovery of the communities in Stochastic Block Model (SBM) is possible in polynomial time above, and only above, the Kesten-Stigum (KS) threshold. This conjecture has given rise to a r...
stat_ml
stat.ML
2025-09-19T09:53:56+00:00
2512.24497v3
What Drives Success in Physical Planning with Joint-Embedding Predictive World Models?
A long-standing challenge in AI is to develop agents capable of solving a wide range of physical tasks and generalizing to new, unseen tasks and environments. A popular recent approach involves training a world model from state-action trajectories and subsequently use it with a planning algorithm to solve new tasks. Pl...
What Drives Success in Physical Planning with Joint-Embedding Predictive World Models?. A long-standing challenge in AI is to develop agents capable of solving a wide range of physical tasks and generalizing to new, unseen tasks and environments. A popular recent approach involves training a world model from state-acti...
ai_general
cs.AI
2025-12-30T22:50:03+00:00
2512.21791v1
Synthetic Financial Data Generation for Enhanced Financial Modelling
Data scarcity and confidentiality in finance often impede model development and robust testing. This paper presents a unified multi-criteria evaluation framework for synthetic financial data and applies it to three representative generative paradigms: the statistical ARIMA-GARCH baseline, Variational Autoencoders (VAEs...
Synthetic Financial Data Generation for Enhanced Financial Modelling. Data scarcity and confidentiality in finance often impede model development and robust testing. This paper presents a unified multi-criteria evaluation framework for synthetic financial data and applies it to three representative generative paradigms...
lg_ml
cs.LG
2025-12-25T21:43:16+00:00
2512.22814v1
Long-Range Distillation: Distilling 10,000 Years of Simulated Climate into Long Timestep AI Weather Models
Accurate long-range weather forecasting remains a major challenge for AI models, both because errors accumulate over autoregressive rollouts and because reanalysis datasets used for training offer a limited sample of the slow modes of climate variability underpinning predictability. Most AI weather models are autoregre...
Long-Range Distillation: Distilling 10,000 Years of Simulated Climate into Long Timestep AI Weather Models. Accurate long-range weather forecasting remains a major challenge for AI models, both because errors accumulate over autoregressive rollouts and because reanalysis datasets used for training offer a limited sampl...
lg_ml
cs.LG
2025-12-28T07:03:20+00:00
2510.01574v1
Synthetic Prefixes to Mitigate Bias in Real-Time Neural Query Autocomplete
We introduce a data-centric approach for mitigating presentation bias in real-time neural query autocomplete systems through the use of synthetic prefixes. These prefixes are generated from complete user queries collected during regular search sessions where autocomplete was not active. This allows us to enrich the tra...
Synthetic Prefixes to Mitigate Bias in Real-Time Neural Query Autocomplete. We introduce a data-centric approach for mitigating presentation bias in real-time neural query autocomplete systems through the use of synthetic prefixes. These prefixes are generated from complete user queries collected during regular search ...
ir_retrieval
cs.IR
2025-10-02T01:44:44+00:00
2512.20762v2
Subgroup Discovery with the Cox Model
We study the problem of subgroup discovery for survival analysis, where the goal is to find an interpretable subset of the data on which a Cox model is highly accurate. Our work is the first to study this particular subgroup problem, for which we make several contributions. Subgroup discovery methods generally require ...
Subgroup Discovery with the Cox Model. We study the problem of subgroup discovery for survival analysis, where the goal is to find an interpretable subset of the data on which a Cox model is highly accurate. Our work is the first to study this particular subgroup problem, for which we make several contributions. Subgro...
lg_ml
cs.LG
2025-12-23T20:49:05+00:00
2512.08743v3
Single-Agent Scaling Fails Multi-Agent Intelligence: Towards Foundation Models with Native Multi-Agent Intelligence
Foundation models (FMs) are increasingly assuming the role of the ''brain'' of AI agents. While recent efforts have begun to equip FMs with native single-agent abilities -- such as GUI interaction or integrated tool use -- we argue that the next frontier is endowing FMs with native multi-agent intelligence. We identify...
Single-Agent Scaling Fails Multi-Agent Intelligence: Towards Foundation Models with Native Multi-Agent Intelligence. Foundation models (FMs) are increasingly assuming the role of the ''brain'' of AI agents. While recent efforts have begun to equip FMs with native single-agent abilities -- such as GUI interaction or int...
ai_general
cs.AI
2025-12-09T15:51:36+00:00
2512.22672v1
Quantum Generative Models for Computational Fluid Dynamics: A First Exploration of Latent Space Learning in Lattice Boltzmann Simulations
This paper presents the first application of quantum generative models to learned latent space representations of computational fluid dynamics (CFD) data. While recent work has explored quantum models for learning statistical properties of fluid systems, the combination of discrete latent space compression with quantum...
Quantum Generative Models for Computational Fluid Dynamics: A First Exploration of Latent Space Learning in Lattice Boltzmann Simulations. This paper presents the first application of quantum generative models to learned latent space representations of computational fluid dynamics (CFD) data. While recent work has expl...
lg_ml
cs.LG
2025-12-27T18:21:09+00:00
2510.12327v1
Simple Projection Variants Improve ColBERT Performance
Multi-vector dense retrieval methods like ColBERT systematically use a single-layer linear projection to reduce the dimensionality of individual vectors. In this study, we explore the implications of the MaxSim operator on the gradient flows of the training of multi-vector models and show that such a simple linear proj...
Simple Projection Variants Improve ColBERT Performance. Multi-vector dense retrieval methods like ColBERT systematically use a single-layer linear projection to reduce the dimensionality of individual vectors. In this study, we explore the implications of the MaxSim operator on the gradient flows of the training of mul...
ir_retrieval
cs.IR
2025-10-14T09:34:05+00:00
2512.14604v1
LLmFPCA-detect: LLM-powered Multivariate Functional PCA for Anomaly Detection in Sparse Longitudinal Texts
Sparse longitudinal (SL) textual data arises when individuals generate text repeatedly over time (e.g., customer reviews, occasional social media posts, electronic medical records across visits), but the frequency and timing of observations vary across individuals. These complex textual data sets have immense potential...
LLmFPCA-detect: LLM-powered Multivariate Functional PCA for Anomaly Detection in Sparse Longitudinal Texts. Sparse longitudinal (SL) textual data arises when individuals generate text repeatedly over time (e.g., customer reviews, occasional social media posts, electronic medical records across visits), but the frequenc...
stat_ml
stat.ML
2025-12-16T17:14:10+00:00
2511.14441v1
Skewness-Robust Causal Discovery in Location-Scale Noise Models
To distinguish Markov equivalent graphs in causal discovery, it is necessary to restrict the structural causal model. Crucially, we need to be able to distinguish cause $X$ from effect $Y$ in bivariate models, that is, distinguish the two graphs $X \to Y$ and $Y \to X$. Location-scale noise models (LSNMs), in which the...
Skewness-Robust Causal Discovery in Location-Scale Noise Models. To distinguish Markov equivalent graphs in causal discovery, it is necessary to restrict the structural causal model. Crucially, we need to be able to distinguish cause $X$ from effect $Y$ in bivariate models, that is, distinguish the two graphs $X \to Y$...
stat_ml
stat.ML
2025-11-18T12:40:41+00:00
2512.17610v1
Semi-Supervised 3D Segmentation for Type-B Aortic Dissection with Slim UNETR
Convolutional neural networks (CNN) for multi-class segmentation of medical images are widely used today. Especially models with multiple outputs that can separately predict segmentation classes (regions) without relying on a probabilistic formulation of the segmentation of regions. These models allow for more precise ...
Semi-Supervised 3D Segmentation for Type-B Aortic Dissection with Slim UNETR. Convolutional neural networks (CNN) for multi-class segmentation of medical images are widely used today. Especially models with multiple outputs that can separately predict segmentation classes (regions) without relying on a probabilistic fo...
cv_vision
cs.CV
2025-12-19T14:14:41+00:00
2512.17041v1
Security Risks of Agentic Vehicles: A Systematic Analysis of Cognitive and Cross-Layer Threats
Agentic AI is increasingly being explored and introduced in both manually driven and autonomous vehicles, leading to the notion of Agentic Vehicles (AgVs), with capabilities such as memory-based personalization, goal interpretation, strategic reasoning, and tool-mediated assistance. While frameworks such as the OWASP A...
Security Risks of Agentic Vehicles: A Systematic Analysis of Cognitive and Cross-Layer Threats. Agentic AI is increasingly being explored and introduced in both manually driven and autonomous vehicles, leading to the notion of Agentic Vehicles (AgVs), with capabilities such as memory-based personalization, goal interpr...
ai_general
cs.AI
2025-12-18T20:04:21+00:00
2510.11895v1
High-Probability Bounds For Heterogeneous Local Differential Privacy
We study statistical estimation under local differential privacy (LDP) when users may hold heterogeneous privacy levels and accuracy must be guaranteed with high probability. Departing from the common in-expectation analyses, and for one-dimensional and multi-dimensional mean estimation problems, we develop finite samp...
High-Probability Bounds For Heterogeneous Local Differential Privacy. We study statistical estimation under local differential privacy (LDP) when users may hold heterogeneous privacy levels and accuracy must be guaranteed with high probability. Departing from the common in-expectation analyses, and for one-dimensional ...
stat_ml
stat.ML
2025-10-13T19:54:44+00:00
2512.15776v1
Emergence: Overcoming Privileged Information Bias in Asymmetric Embodied Agents via Active Querying
Large Language Models (LLMs) act as powerful reasoning engines but struggle with "symbol grounding" in embodied environments, particularly when information is asymmetrically distributed. We investigate the Privileged Information Bias (or "Curse of Knowledge"), where a knowledgeable "Leader" agent fails to guide a senso...
Emergence: Overcoming Privileged Information Bias in Asymmetric Embodied Agents via Active Querying. Large Language Models (LLMs) act as powerful reasoning engines but struggle with "symbol grounding" in embodied environments, particularly when information is asymmetrically distributed. We investigate the Privileged In...
ai_general
cs.AI
2025-12-13T17:17:51+00:00
2512.20520v1
Benchmarking LLMs for Predictive Applications in the Intensive Care Units
With the advent of LLMs, various tasks across the natural language processing domain have been transformed. However, their application in predictive tasks remains less researched. This study compares large language models, including GatorTron-Base (trained on clinical data), Llama 8B, and Mistral 7B, against models lik...
Benchmarking LLMs for Predictive Applications in the Intensive Care Units. With the advent of LLMs, various tasks across the natural language processing domain have been transformed. However, their application in predictive tasks remains less researched. This study compares large language models, including GatorTron-Ba...
ai_general
cs.AI
2025-12-23T17:08:31+00:00
2509.02266v1
Leveraging Media Frames to Improve Normative Diversity in News Recommendations
Click-based news recommender systems suggest users content that aligns with their existing history, limiting the diversity of articles they encounter. Recent advances in aspect-based diversification -- adding features such as sentiments or news categories (e.g. world, politics) -- have made progress toward diversifying...
Leveraging Media Frames to Improve Normative Diversity in News Recommendations. Click-based news recommender systems suggest users content that aligns with their existing history, limiting the diversity of articles they encounter. Recent advances in aspect-based diversification -- adding features such as sentiments or ...
ir_retrieval
cs.IR
2025-09-02T12:43:41+00:00
2509.25126v1
On Spectral Learning for Odeco Tensors: Perturbation, Initialization, and Algorithms
We study spectral learning for orthogonally decomposable (odeco) tensors, emphasizing the interplay between statistical limits, optimization geometry, and initialization. Unlike matrices, recovery for odeco tensors does not hinge on eigengaps, yielding improved robustness under noise. While iterative methods such as te...
On Spectral Learning for Odeco Tensors: Perturbation, Initialization, and Algorithms. We study spectral learning for orthogonally decomposable (odeco) tensors, emphasizing the interplay between statistical limits, optimization geometry, and initialization. Unlike matrices, recovery for odeco tensors does not hinge on e...
stat_ml
stat.ML
2025-09-29T17:45:35+00:00
2510.19372v2
On the Hardness of Reinforcement Learning with Transition Look-Ahead
We study reinforcement learning (RL) with transition look-ahead, where the agent may observe which states would be visited upon playing any sequence of $\ell$ actions before deciding its course of action. While such predictive information can drastically improve the achievable performance, we show that using this infor...
On the Hardness of Reinforcement Learning with Transition Look-Ahead. We study reinforcement learning (RL) with transition look-ahead, where the agent may observe which states would be visited upon playing any sequence of $\ell$ actions before deciding its course of action. While such predictive information can drastic...
stat_ml
stat.ML
2025-10-22T08:47:18+00:00
2512.24591v1
Improving Few-Shot Change Detection Visual Question Answering via Decision-Ambiguity-guided Reinforcement Fine-Tuning
Change detection visual question answering (CDVQA) requires answering text queries by reasoning about semantic changes in bi-temporal remote sensing images. A straightforward approach is to boost CDVQA performance with generic vision-language models via supervised fine-tuning (SFT). Despite recent progress, we observe ...
Improving Few-Shot Change Detection Visual Question Answering via Decision-Ambiguity-guided Reinforcement Fine-Tuning. Change detection visual question answering (CDVQA) requires answering text queries by reasoning about semantic changes in bi-temporal remote sensing images. A straightforward approach is to boost CDVQA...
cv_vision
cs.CV
2025-12-31T03:28:17+00:00
2512.22857v1
AutoForge: Automated Environment Synthesis for Agentic Reinforcement Learning
Conducting reinforcement learning (RL) in simulated environments offers a cost-effective and highly scalable way to enhance language-based agents. However, previous work has been limited to semi-automated environment synthesis or tasks lacking sufficient difficulty, offering little breadth or depth. In addition, the in...
AutoForge: Automated Environment Synthesis for Agentic Reinforcement Learning. Conducting reinforcement learning (RL) in simulated environments offers a cost-effective and highly scalable way to enhance language-based agents. However, previous work has been limited to semi-automated environment synthesis or tasks lacki...
cl_nlp
cs.CL
2025-12-28T09:43:11+00:00
2512.16532v1
From Personalization to Prejudice: Bias and Discrimination in Memory-Enhanced AI Agents for Recruitment
Large Language Models (LLMs) have empowered AI agents with advanced capabilities for understanding, reasoning, and interacting across diverse tasks. The addition of memory further enhances them by enabling continuity across interactions, learning from past experiences, and improving the relevance of actions and respons...
From Personalization to Prejudice: Bias and Discrimination in Memory-Enhanced AI Agents for Recruitment. Large Language Models (LLMs) have empowered AI agents with advanced capabilities for understanding, reasoning, and interacting across diverse tasks. The addition of memory further enhances them by enabling continuit...
ai_general
cs.AI
2025-12-18T13:41:37+00:00
2512.21053v1
Optical Flow-Guided 6DoF Object Pose Tracking with an Event Camera
Object pose tracking is one of the pivotal technologies in multimedia, attracting ever-growing attention in recent years. Existing methods employing traditional cameras encounter numerous challenges such as motion blur, sensor noise, partial occlusion, and changing lighting conditions. The emerging bio-inspired sensors...
Optical Flow-Guided 6DoF Object Pose Tracking with an Event Camera. Object pose tracking is one of the pivotal technologies in multimedia, attracting ever-growing attention in recent years. Existing methods employing traditional cameras encounter numerous challenges such as motion blur, sensor noise, partial occlusion,...
cv_vision
cs.CV
2025-12-24T08:40:57+00:00
2510.08109v1
VersionRAG: Version-Aware Retrieval-Augmented Generation for Evolving Documents
Retrieval-Augmented Generation (RAG) systems fail when documents evolve through versioning-a ubiquitous characteristic of technical documentation. Existing approaches achieve only 58-64% accuracy on version-sensitive questions, retrieving semantically similar content without temporal validity checks. We present Version...
VersionRAG: Version-Aware Retrieval-Augmented Generation for Evolving Documents. Retrieval-Augmented Generation (RAG) systems fail when documents evolve through versioning-a ubiquitous characteristic of technical documentation. Existing approaches achieve only 58-64% accuracy on version-sensitive questions, retrieving ...
ir_retrieval
cs.IR
2025-10-09T11:48:58+00:00
2512.24165v1
DiffThinker: Towards Generative Multimodal Reasoning with Diffusion Models
While recent Multimodal Large Language Models (MLLMs) have attained significant strides in multimodal reasoning, their reasoning processes remain predominantly text-centric, leading to suboptimal performance in complex long-horizon, vision-centric tasks. In this paper, we establish a novel Generative Multimodal Reasoni...
DiffThinker: Towards Generative Multimodal Reasoning with Diffusion Models. While recent Multimodal Large Language Models (MLLMs) have attained significant strides in multimodal reasoning, their reasoning processes remain predominantly text-centric, leading to suboptimal performance in complex long-horizon, vision-cent...
cv_vision
cs.CV
2025-12-30T11:51:18+00:00
2512.18331v1
A two-stream network with global-local feature fusion for bone age assessment
Bone Age Assessment (BAA) is a widely used clinical technique that can accurately reflect an individual's growth and development level, as well as maturity. In recent years, although deep learning has advanced the field of bone age assessment, existing methods face challenges in efficiently balancing global features an...
A two-stream network with global-local feature fusion for bone age assessment. Bone Age Assessment (BAA) is a widely used clinical technique that can accurately reflect an individual's growth and development level, as well as maturity. In recent years, although deep learning has advanced the field of bone age assessmen...
cv_vision
cs.CV
2025-12-20T11:56:21+00:00
2602.22224v1
DS SERVE: A Framework for Efficient and Scalable Neural Retrieval
We present DS-Serve, a framework that transforms large-scale text datasets, comprising half a trillion tokens, into a high-performance neural retrieval system. DS-Serve offers both a web interface and API endpoints, achieving low latency with modest memory overhead on a single node. The framework also supports inferenc...
DS SERVE: A Framework for Efficient and Scalable Neural Retrieval. We present DS-Serve, a framework that transforms large-scale text datasets, comprising half a trillion tokens, into a high-performance neural retrieval system. DS-Serve offers both a web interface and API endpoints, achieving low latency with modest mem...
ir_retrieval
cs.IR
2025-12-17T00:43:10+00:00
2512.17459v1
3D-RE-GEN: 3D Reconstruction of Indoor Scenes with a Generative Framework
Recent advances in 3D scene generation produce visually appealing output, but current representations hinder artists' workflows that require modifiable 3D textured mesh scenes for visual effects and game development. Despite significant advances, current textured mesh scene reconstruction methods are far from artist re...
3D-RE-GEN: 3D Reconstruction of Indoor Scenes with a Generative Framework. Recent advances in 3D scene generation produce visually appealing output, but current representations hinder artists' workflows that require modifiable 3D textured mesh scenes for visual effects and game development. Despite significant advances...
cv_vision
cs.CV
2025-12-19T11:20:52+00:00
2512.12503v1
KidsArtBench: Multi-Dimensional Children's Art Evaluation with Attribute-Aware MLLMs
Multimodal Large Language Models (MLLMs) show remarkable progress across many visual-language tasks; however, their capacity to evaluate artistic expression remains limited. Aesthetic concepts are inherently abstract and open-ended, and multimodal artwork annotations are scarce. We introduce KidsArtBench, a new benchma...
KidsArtBench: Multi-Dimensional Children's Art Evaluation with Attribute-Aware MLLMs. Multimodal Large Language Models (MLLMs) show remarkable progress across many visual-language tasks; however, their capacity to evaluate artistic expression remains limited. Aesthetic concepts are inherently abstract and open-ended, a...
ai_general
cs.AI
2025-12-14T00:24:48+00:00
2512.04408v1
Executable Governance for AI: Translating Policies into Rules Using LLMs
AI policy guidance is predominantly written as prose, which practitioners must first convert into executable rules before frameworks can evaluate or enforce them. This manual step is slow, error-prone, difficult to scale, and often delays the use of safeguards in real-world deployments. To address this gap, we present ...
Executable Governance for AI: Translating Policies into Rules Using LLMs. AI policy guidance is predominantly written as prose, which practitioners must first convert into executable rules before frameworks can evaluate or enforce them. This manual step is slow, error-prone, difficult to scale, and often delays the use...
ai_general
cs.AI
2025-12-04T03:11:54+00:00
2512.22009v1
iSHIFT: Lightweight Slow-Fast GUI Agent with Adaptive Perception
Multimodal Large Language Models (MLLMs) show strong potential for interpreting and interacting with complex, pixel-rich Graphical User Interface (GUI) environments. However, building agents that are both efficient for high-level tasks and precise for fine-grained interactions remains challenging. GUI agents must perfo...
iSHIFT: Lightweight Slow-Fast GUI Agent with Adaptive Perception. Multimodal Large Language Models (MLLMs) show strong potential for interpreting and interacting with complex, pixel-rich Graphical User Interface (GUI) environments. However, building agents that are both efficient for high-level tasks and precise for fi...
cv_vision
cs.CV
2025-12-26T12:09:15+00:00
2512.05414v3
LMSpell: Neural Spell Checking for Low-Resource Languages
Spell correction is still a challenging problem for low-resource languages (LRLs). While pretrained language models (PLMs) have been employed for spell correction, their use is still limited to a handful of languages, and there has been no proper comparison across PLMs. We present the first empirical study on the effec...
LMSpell: Neural Spell Checking for Low-Resource Languages. Spell correction is still a challenging problem for low-resource languages (LRLs). While pretrained language models (PLMs) have been employed for spell correction, their use is still limited to a handful of languages, and there has been no proper comparison acr...
cl_nlp
cs.CL
2025-12-05T04:14:09+00:00
2510.20653v1
Finding the Sweet Spot: Trading Quality, Cost, and Speed During Inference-Time LLM Reflection
As Large Language Models (LLMs) continue to evolve, practitioners face increasing options for enhancing inference-time performance without model retraining, including budget tuning and multi-step techniques like self-reflection. While these methods improve output quality, they create complex trade-offs among accuracy, ...
Finding the Sweet Spot: Trading Quality, Cost, and Speed During Inference-Time LLM Reflection. As Large Language Models (LLMs) continue to evolve, practitioners face increasing options for enhancing inference-time performance without model retraining, including budget tuning and multi-step techniques like self-reflecti...
stat_ml
stat.ML
2025-10-23T15:26:18+00:00
2509.06576v1
Automated Hierarchical Graph Construction for Multi-source Electronic Health Records
Electronic Health Records (EHRs), comprising diverse clinical data such as diagnoses, medications, and laboratory results, hold great promise for translational research. EHR-derived data have advanced disease prevention, improved clinical trial recruitment, and generated real-world evidence. Synthesizing EHRs across in...
Automated Hierarchical Graph Construction for Multi-source Electronic Health Records. Electronic Health Records (EHRs), comprising diverse clinical data such as diagnoses, medications, and laboratory results, hold great promise for translational research. EHR-derived data have advanced disease prevention, improved clin...
stat_ml
stat.ML
2025-09-08T11:45:59+00:00
2512.10821v2
Agile Deliberation: Concept Deliberation for Subjective Visual Classification
From content moderation to content curation, applications requiring vision classifiers for visual concepts are rapidly expanding. Existing human-in-the-loop approaches typically assume users begin with a clear, stable concept understanding to be able to provide high-quality supervision. In reality, users often start wi...
Agile Deliberation: Concept Deliberation for Subjective Visual Classification. From content moderation to content curation, applications requiring vision classifiers for visual concepts are rapidly expanding. Existing human-in-the-loop approaches typically assume users begin with a clear, stable concept understanding t...
ai_general
cs.AI
2025-12-11T17:13:09+00:00
2509.20228v1
Muse-it: A Tool for Analyzing Music Discourse on Reddit
Music engagement spans diverse interactions with music, from selection and emotional response to its impact on behavior, identity, and social connections. Social media platforms provide spaces where such engagement can be observed in natural, unprompted conversations. Advances in natural language processing (NLP) and b...
Muse-it: A Tool for Analyzing Music Discourse on Reddit. Music engagement spans diverse interactions with music, from selection and emotional response to its impact on behavior, identity, and social connections. Social media platforms provide spaces where such engagement can be observed in natural, unprompted conversat...
ir_retrieval
cs.IR
2025-09-24T15:22:23+00:00
2512.21584v1
UltraLBM-UNet: Ultralight Bidirectional Mamba-based Model for Skin Lesion Segmentation
Skin lesion segmentation is a crucial step in dermatology for guiding clinical decision-making. However, existing methods for accurate, robust, and resource-efficient lesion analysis have limitations, including low performance and high computational complexity. To address these limitations, we propose UltraLBM-UNet, a ...
UltraLBM-UNet: Ultralight Bidirectional Mamba-based Model for Skin Lesion Segmentation. Skin lesion segmentation is a crucial step in dermatology for guiding clinical decision-making. However, existing methods for accurate, robust, and resource-efficient lesion analysis have limitations, including low performance and h...
cv_vision
cs.CV
2025-12-25T09:05:02+00:00
2510.01291v1
Private Realizable-to-Agnostic Transformation with Near-Optimal Sample Complexity
The realizable-to-agnostic transformation (Beimel et al., 2015; Alon et al., 2020) provides a general mechanism to convert a private learner in the realizable setting (where the examples are labeled by some function in the concept class) to a private learner in the agnostic setting (where no assumptions are imposed on ...
Private Realizable-to-Agnostic Transformation with Near-Optimal Sample Complexity. The realizable-to-agnostic transformation (Beimel et al., 2015; Alon et al., 2020) provides a general mechanism to convert a private learner in the realizable setting (where the examples are labeled by some function in the concept class)...
stat_ml
stat.ML
2025-10-01T04:49:43+00:00
2512.23042v2
3D sans 3D Scans: Scalable Pre-training from Video-Generated Point Clouds
Despite recent progress in 3D self-supervised learning, collecting large-scale 3D scene scans remains expensive and labor-intensive. In this work, we investigate whether 3D representations can be learned from unlabeled videos recorded without any real 3D sensors. We present Laplacian-Aware Multi-level 3D Clustering wit...
3D sans 3D Scans: Scalable Pre-training from Video-Generated Point Clouds. Despite recent progress in 3D self-supervised learning, collecting large-scale 3D scene scans remains expensive and labor-intensive. In this work, we investigate whether 3D representations can be learned from unlabeled videos recorded without an...
cv_vision
cs.CV
2025-12-28T18:59:25+00:00
2510.06924v1
Ethical AI prompt recommendations in large language models using collaborative filtering
As large language models (LLMs) shape AI development, ensuring ethical prompt recommendations is crucial. LLMs offer innovation but risk bias, fairness issues, and accountability concerns. Traditional oversight methods struggle with scalability, necessitating dynamic solutions. This paper proposes using collaborative f...
Ethical AI prompt recommendations in large language models using collaborative filtering. As large language models (LLMs) shape AI development, ensuring ethical prompt recommendations is crucial. LLMs offer innovation but risk bias, fairness issues, and accountability concerns. Traditional oversight methods struggle wi...
ir_retrieval
cs.IR
2025-10-08T12:03:21+00:00
2512.22725v1
Mitigating Social Desirability Bias in Random Silicon Sampling
Large Language Models (LLMs) are increasingly used to simulate population responses, a method known as ``Silicon Sampling''. However, responses to socially sensitive questions frequently exhibit Social Desirability Bias (SDB), diverging from real human data toward socially acceptable answers. Existing studies on social...
Mitigating Social Desirability Bias in Random Silicon Sampling. Large Language Models (LLMs) are increasingly used to simulate population responses, a method known as ``Silicon Sampling''. However, responses to socially sensitive questions frequently exhibit Social Desirability Bias (SDB), diverging from real human dat...
cl_nlp
cs.CL
2025-12-27T23:21:32+00:00
2512.22872v2
Lamps: Learning Anatomy from Multiple Perspectives via Self-supervision in Chest Radiographs
Foundation models have been successful in natural language processing and computer vision because they are capable of capturing the underlying structures (foundation) of natural languages. However, in medical imaging, the key foundation lies in human anatomy, as these images directly represent the internal structures o...
Lamps: Learning Anatomy from Multiple Perspectives via Self-supervision in Chest Radiographs. Foundation models have been successful in natural language processing and computer vision because they are capable of capturing the underlying structures (foundation) of natural languages. However, in medical imaging, the key ...
cv_vision
cs.CV
2025-12-28T10:52:51+00:00
2512.18266v1
LeJOT: An Intelligent Job Cost Orchestration Solution for Databricks Platform
With the rapid advancements in big data technologies, the Databricks platform has become a cornerstone for enterprises and research institutions, offering high computational efficiency and a robust ecosystem. However, managing the escalating operational costs associated with job execution remains a critical challenge. ...
LeJOT: An Intelligent Job Cost Orchestration Solution for Databricks Platform. With the rapid advancements in big data technologies, the Databricks platform has become a cornerstone for enterprises and research institutions, offering high computational efficiency and a robust ecosystem. However, managing the escalating...
lg_ml
cs.LG
2025-12-20T08:09:58+00:00
2510.05624v1
Limitations of Current Evaluation Practices for Conversational Recommender Systems and the Potential of User Simulation
Research and development on conversational recommender systems (CRSs) critically depends on sound and reliable evaluation methodologies. However, the interactive nature of these systems poses significant challenges for automatic evaluation. This paper critically examines current evaluation practices and identifies two ...
Limitations of Current Evaluation Practices for Conversational Recommender Systems and the Potential of User Simulation. Research and development on conversational recommender systems (CRSs) critically depends on sound and reliable evaluation methodologies. However, the interactive nature of these systems poses signifi...
ir_retrieval
cs.IR
2025-10-07T07:12:47+00:00
2512.23472v1
MCI-Net: A Robust Multi-Domain Context Integration Network for Point Cloud Registration
Robust and discriminative feature learning is critical for high-quality point cloud registration. However, existing deep learning-based methods typically rely on Euclidean neighborhood-based strategies for feature extraction, which struggle to effectively capture the implicit semantics and structural consistency in poi...
MCI-Net: A Robust Multi-Domain Context Integration Network for Point Cloud Registration. Robust and discriminative feature learning is critical for high-quality point cloud registration. However, existing deep learning-based methods typically rely on Euclidean neighborhood-based strategies for feature extraction, which...
cv_vision
cs.CV
2025-12-29T13:55:33+00:00
2512.24826v1
Video and Language Alignment in 2D Systems for 3D Multi-object Scenes with Multi-Information Derivative-Free Control
Cross-modal systems trained on 2D visual inputs are presented with a dimensional shift when processing 3D scenes. An in-scene camera bridges the dimensionality gap but requires learning a control module. We introduce a new method that improves multivariate mutual information estimates by regret minimisation with deriva...
Video and Language Alignment in 2D Systems for 3D Multi-object Scenes with Multi-Information Derivative-Free Control. Cross-modal systems trained on 2D visual inputs are presented with a dimensional shift when processing 3D scenes. An in-scene camera bridges the dimensionality gap but requires learning a control module...
cv_vision
cs.CV
2025-12-31T12:39:03+00:00
2512.21643v3
Omni-Weather: A Unified Multimodal Model for Weather Radar Understanding and Generation
Weather modeling requires both accurate prediction and mechanistic interpretation, yet existing methods treat these goals in isolation, separating generation from understanding. To address this gap, we present Omni-Weather, the first multimodal foundation model that unifies weather generation and understanding within a...
Omni-Weather: A Unified Multimodal Model for Weather Radar Understanding and Generation. Weather modeling requires both accurate prediction and mechanistic interpretation, yet existing methods treat these goals in isolation, separating generation from understanding. To address this gap, we present Omni-Weather, the fir...
cv_vision
cs.CV
2025-12-25T12:08:09+00:00
2512.19804v1
Reduced Order Modeling for Tsunami Forecasting with Bayesian Hierarchical Pooling
Reduced order models (ROM) can represent spatiotemporal processes in significantly fewer dimensions and can be solved many orders faster than their governing partial differential equations (PDEs). For example, using a proper orthogonal decomposition produces a ROM that is a small linear combination of fixed features an...
Reduced Order Modeling for Tsunami Forecasting with Bayesian Hierarchical Pooling. Reduced order models (ROM) can represent spatiotemporal processes in significantly fewer dimensions and can be solved many orders faster than their governing partial differential equations (PDEs). For example, using a proper orthogonal d...
lg_ml
cs.LG
2025-12-22T19:01:42+00:00
2512.03234v1
Iterative Tilting for Diffusion Fine-Tuning
We introduce iterative tilting, a gradient-free method for fine-tuning diffusion models toward reward-tilted distributions. The method decomposes a large reward tilt $\exp(λr)$ into $N$ sequential smaller tilts, each admitting a tractable score update via first-order Taylor expansion. This requires only forward evaluat...
Iterative Tilting for Diffusion Fine-Tuning. We introduce iterative tilting, a gradient-free method for fine-tuning diffusion models toward reward-tilted distributions. The method decomposes a large reward tilt $\exp(λr)$ into $N$ sequential smaller tilts, each admitting a tractable score update via first-order Taylor ...
stat_ml
stat.ML
2025-12-02T21:07:46+00:00
2512.20777v2
Improving Matrix Exponential for Generative AI Flows: A Taylor-Based Approach Beyond Paterson--Stockmeyer
The matrix exponential is a fundamental operator in scientific computing and system simulation, with applications ranging from control theory and quantum mechanics to modern generative machine learning. While Padé approximants combined with scaling and squaring have long served as the standard, recent Taylor-based meth...
Improving Matrix Exponential for Generative AI Flows: A Taylor-Based Approach Beyond Paterson--Stockmeyer. The matrix exponential is a fundamental operator in scientific computing and system simulation, with applications ranging from control theory and quantum mechanics to modern generative machine learning. While Padé...
lg_ml
cs.LG
2025-12-23T21:25:40+00:00
2511.15141v2
ItemRAG: Item-Based Retrieval-Augmented Generation for LLM-Based Recommendation
Recently, large language models (LLMs) have been widely used as recommender systems, owing to their reasoning capability and effectiveness in handling cold-start items. A common approach prompts an LLM with a target user's purchase history to recommend items from a candidate set, often enhanced with retrieval-augmented...
ItemRAG: Item-Based Retrieval-Augmented Generation for LLM-Based Recommendation. Recently, large language models (LLMs) have been widely used as recommender systems, owing to their reasoning capability and effectiveness in handling cold-start items. A common approach prompts an LLM with a target user's purchase history...
ir_retrieval
cs.IR
2025-11-19T05:39:14+00:00
2511.19404v1
Nonparametric Instrumental Variable Regression with Observed Covariates
We study the problem of nonparametric instrumental variable regression with observed covariates, which we refer to as NPIV-O. Compared with standard nonparametric instrumental variable regression (NPIV), the additional observed covariates facilitate causal identification and enables heterogeneous causal effect estimati...
Nonparametric Instrumental Variable Regression with Observed Covariates. We study the problem of nonparametric instrumental variable regression with observed covariates, which we refer to as NPIV-O. Compared with standard nonparametric instrumental variable regression (NPIV), the additional observed covariates facilita...
stat_ml
stat.ML
2025-11-24T18:42:49+00:00
2512.17376v1
Towards Deeper Emotional Reflection: Crafting Affective Image Filters with Generative Priors
Social media platforms enable users to express emotions by posting text with accompanying images. In this paper, we propose the Affective Image Filter (AIF) task, which aims to reflect visually-abstract emotions from text into visually-concrete images, thereby creating emotionally compelling results. We first introduce...
Towards Deeper Emotional Reflection: Crafting Affective Image Filters with Generative Priors. Social media platforms enable users to express emotions by posting text with accompanying images. In this paper, we propose the Affective Image Filter (AIF) task, which aims to reflect visually-abstract emotions from text into...
cv_vision
cs.CV
2025-12-19T09:24:22+00:00
2512.16442v1
TIB AIssistant: a Platform for AI-Supported Research Across Research Life Cycles
The rapidly growing popularity of adopting Artificial Intelligence (AI), and specifically Large Language Models (LLMs), is having a widespread impact throughout society, including the academic domain. AI-supported research has the potential to support researchers with tasks across the entire research life cycle. In thi...
TIB AIssistant: a Platform for AI-Supported Research Across Research Life Cycles. The rapidly growing popularity of adopting Artificial Intelligence (AI), and specifically Large Language Models (LLMs), is having a widespread impact throughout society, including the academic domain. AI-supported research has the potenti...
ai_general
cs.AI
2025-12-18T11:54:38+00:00
2512.21699v1
Towards Responsible and Explainable AI Agents with Consensus-Driven Reasoning
Agentic AI represents a major shift in how autonomous systems reason, plan, and execute multi-step tasks through the coordination of Large Language Models (LLMs), Vision Language Models (VLMs), tools, and external services. While these systems enable powerful new capabilities, increasing autonomy introduces critical ch...
Towards Responsible and Explainable AI Agents with Consensus-Driven Reasoning. Agentic AI represents a major shift in how autonomous systems reason, plan, and execute multi-step tasks through the coordination of Large Language Models (LLMs), Vision Language Models (VLMs), tools, and external services. While these syste...
ai_general
cs.AI
2025-12-25T14:49:25+00:00
2512.23578v3
Style Amnesia: Investigating Speaking Style Degradation and Mitigation in Multi-Turn Spoken Language Models
In this paper, we show that when spoken language models (SLMs) are instructed to speak in a specific speaking style at the beginning of a multi-turn conversation, they cannot maintain the required speaking styles after several turns of interaction; we refer to this as the style amnesia of SLMs. We focus on paralinguist...
Style Amnesia: Investigating Speaking Style Degradation and Mitigation in Multi-Turn Spoken Language Models. In this paper, we show that when spoken language models (SLMs) are instructed to speak in a specific speaking style at the beginning of a multi-turn conversation, they cannot maintain the required speaking style...
cl_nlp
cs.CL
2025-12-29T16:23:54+00:00
2512.18466v1
Self-organizing maps for water quality assessment in reservoirs and lakes: A systematic literature review
Sustainable water quality underpins ecological balance and water security. Assessing and managing lakes and reservoirs is difficult due to data sparsity, heterogeneity, and nonlinear relationships among parameters. This review examines how Self-Organizing Map (SOM), an unsupervised AI technique, is applied to water qua...
Self-organizing maps for water quality assessment in reservoirs and lakes: A systematic literature review. Sustainable water quality underpins ecological balance and water security. Assessing and managing lakes and reservoirs is difficult due to data sparsity, heterogeneity, and nonlinear relationships among parameters...
lg_ml
cs.LG
2025-12-20T18:48:33+00:00
2509.19876v1
Adaptive User Interest Modeling via Conditioned Denoising Diffusion For Click-Through Rate Prediction
User behavior sequences in search systems resemble "interest fossils", capturing genuine intent yet eroded by exposure bias, category drift, and contextual noise. Current methods predominantly follow an "identify-aggregate" paradigm, assuming sequences immutably reflect user preferences while overlooking the organic en...
Adaptive User Interest Modeling via Conditioned Denoising Diffusion For Click-Through Rate Prediction. User behavior sequences in search systems resemble "interest fossils", capturing genuine intent yet eroded by exposure bias, category drift, and contextual noise. Current methods predominantly follow an "identify-aggr...
ir_retrieval
cs.IR
2025-09-24T08:28:33+00:00
2512.13059v1
An Open and Reproducible Deep Research Agent for Long-Form Question Answering
We present an open deep research system for long-form question answering, selected as a winning system in the text-to-text track of the MMU-RAG competition at NeurIPS 2025. The system combines an open-source large language model (LLM) with an open web search API to perform iterative retrieval, reasoning, and synthesis ...
An Open and Reproducible Deep Research Agent for Long-Form Question Answering. We present an open deep research system for long-form question answering, selected as a winning system in the text-to-text track of the MMU-RAG competition at NeurIPS 2025. The system combines an open-source large language model (LLM) with a...
cl_nlp
cs.CL
2025-12-15T07:37:53+00:00
2511.00849v1
Perturbations in the Orthogonal Complement Subspace for Efficient Out-of-Distribution Detection
Out-of-distribution (OOD) detection is essential for deploying deep learning models in open-world environments. Existing approaches, such as energy-based scoring and gradient-projection methods, typically rely on high-dimensional representations to separate in-distribution (ID) and OOD samples. We introduce P-OCS (Pert...
Perturbations in the Orthogonal Complement Subspace for Efficient Out-of-Distribution Detection. Out-of-distribution (OOD) detection is essential for deploying deep learning models in open-world environments. Existing approaches, such as energy-based scoring and gradient-projection methods, typically rely on high-dimen...
stat_ml
stat.ML
2025-11-02T08:21:13+00:00
2510.02657v3
Less LLM, More Documents: Searching for Improved RAG
Retrieval-Augmented Generation (RAG) couples document retrieval with large language models (LLMs). While scaling generators often improves accuracy, it also increases inference and deployment overhead. We study an orthogonal axis: enlarging the retriever's corpus, and how it trades off with generator scale. Across mult...
Less LLM, More Documents: Searching for Improved RAG. Retrieval-Augmented Generation (RAG) couples document retrieval with large language models (LLMs). While scaling generators often improves accuracy, it also increases inference and deployment overhead. We study an orthogonal axis: enlarging the retriever's corpus, a...
ir_retrieval
cs.IR
2025-10-03T01:26:13+00:00
2512.20173v1
Offline Safe Policy Optimization From Heterogeneous Feedback
Offline Preference-based Reinforcement Learning (PbRL) learns rewards and policies aligned with human preferences without the need for extensive reward engineering and direct interaction with human annotators. However, ensuring safety remains a critical challenge across many domains and tasks. Previous works on safe RL...
Offline Safe Policy Optimization From Heterogeneous Feedback. Offline Preference-based Reinforcement Learning (PbRL) learns rewards and policies aligned with human preferences without the need for extensive reward engineering and direct interaction with human annotators. However, ensuring safety remains a critical chal...
ai_general
cs.AI
2025-12-23T09:07:53+00:00
2509.25599v2
Coupling Generative Modeling and an Autoencoder with the Causal Bridge
We consider inferring the causal effect of a treatment (intervention) on an outcome of interest in situations where there is potentially an unobserved confounder influencing both the treatment and the outcome. This is achievable by assuming access to two separate sets of control (proxy) measurements associated with tre...
Coupling Generative Modeling and an Autoencoder with the Causal Bridge. We consider inferring the causal effect of a treatment (intervention) on an outcome of interest in situations where there is potentially an unobserved confounder influencing both the treatment and the outcome. This is achievable by assuming access ...
stat_ml
stat.ML
2025-09-29T23:46:54+00:00
2510.21711v1
Improving E-commerce Search with Category-Aligned Retrieval
Traditional e-commerce search systems often struggle with the semantic gap between user queries and product catalogs. In this paper, we propose a Category-Aligned Retrieval System (CARS) that improves search relevance by first predicting the product category from a user's query and then boosting products within that ca...
Improving E-commerce Search with Category-Aligned Retrieval. Traditional e-commerce search systems often struggle with the semantic gap between user queries and product catalogs. In this paper, we propose a Category-Aligned Retrieval System (CARS) that improves search relevance by first predicting the product category ...
ir_retrieval
cs.IR
2025-09-03T20:43:52+00:00
2512.21005v2
Learning from Neighbors with PHIBP: Predicting Infectious Disease Dynamics in Data-Sparse Environments
Modeling sparse count data, which arise across numerous scientific fields, presents significant statistical challenges. This chapter addresses these challenges in the context of infectious disease prediction, with a focus on predicting outbreaks in geographic regions that have historically reported zero cases. To this ...
Learning from Neighbors with PHIBP: Predicting Infectious Disease Dynamics in Data-Sparse Environments. Modeling sparse count data, which arise across numerous scientific fields, presents significant statistical challenges. This chapter addresses these challenges in the context of infectious disease prediction, with a ...
stat_ml
stat.ML
2025-12-24T07:10:17+00:00
2512.18908v2
Multimodal Bayesian Network for Robust Assessment of Casualties in Autonomous Triage
Mass Casualty Incidents can overwhelm emergency medical systems and resulting delays or errors in the assessment of casualties can lead to preventable deaths. We present a decision support framework that fuses outputs from multiple computer vision models, estimating signs of severe hemorrhage, respiratory distress, phy...
Multimodal Bayesian Network for Robust Assessment of Casualties in Autonomous Triage. Mass Casualty Incidents can overwhelm emergency medical systems and resulting delays or errors in the assessment of casualties can lead to preventable deaths. We present a decision support framework that fuses outputs from multiple co...
ai_general
cs.AI
2025-12-21T22:59:58+00:00
2512.23805v3
Fitted $Q$ Evaluation Without Bellman Completeness via Stationary Weighting
Fitted $Q$-evaluation (FQE) is a standard regression-based tool for off-policy evaluation, but existing stability guarantees often rely on Bellman completeness, a strong closure condition that can fail under function approximation. We study an alternative route: changing the norm used in the regression step. The policy...
Fitted $Q$ Evaluation Without Bellman Completeness via Stationary Weighting. Fitted $Q$-evaluation (FQE) is a standard regression-based tool for off-policy evaluation, but existing stability guarantees often rely on Bellman completeness, a strong closure condition that can fail under function approximation. We study an...
stat_ml
stat.ML
2025-12-29T19:04:40+00:00
2512.05257v1
Resolving Zadehs Paradox Axiomatic Possibility Theory as a Foundation for Reliable Artificial Intelligence
This work advances and substantiates the thesis that the resolution of this crisis lies in the domain of possibility theory, specifically in the axiomatic approach developed in Bychkovs article. Unlike numerous attempts to fix Dempster rule, this approach builds from scratch a logically consistent and mathematically ri...
Resolving Zadehs Paradox Axiomatic Possibility Theory as a Foundation for Reliable Artificial Intelligence. This work advances and substantiates the thesis that the resolution of this crisis lies in the domain of possibility theory, specifically in the axiomatic approach developed in Bychkovs article. Unlike numerous a...
ai_general
cs.AI
2025-12-04T21:13:16+00:00
2512.19361v1
Interpretable Hybrid Deep Q-Learning Framework for IoT-Based Food Spoilage Prediction with Synthetic Data Generation and Hardware Validation
The need for an intelligent, real-time spoilage prediction system has become critical in modern IoT-driven food supply chains, where perishable goods are highly susceptible to environmental conditions. Existing methods often lack adaptability to dynamic conditions and fail to optimize decision making in real time. To a...
Interpretable Hybrid Deep Q-Learning Framework for IoT-Based Food Spoilage Prediction with Synthetic Data Generation and Hardware Validation. The need for an intelligent, real-time spoilage prediction system has become critical in modern IoT-driven food supply chains, where perishable goods are highly susceptible to en...
lg_ml
cs.LG
2025-12-22T12:59:48+00:00
2512.20618v1
LongVideoAgent: Multi-Agent Reasoning with Long Videos
Recent advances in multimodal LLMs and systems that use tools for long-video QA point to the promise of reasoning over hour-long episodes. However, many methods still compress content into lossy summaries or rely on limited toolsets, weakening temporal grounding and missing fine-grained cues. We propose a multi-agent f...
LongVideoAgent: Multi-Agent Reasoning with Long Videos. Recent advances in multimodal LLMs and systems that use tools for long-video QA point to the promise of reasoning over hour-long episodes. However, many methods still compress content into lossy summaries or rely on limited toolsets, weakening temporal grounding a...
ai_general
cs.AI
2025-12-23T18:59:49+00:00
2512.11920v1
CXL-SpecKV: A Disaggregated FPGA Speculative KV-Cache for Datacenter LLM Serving
Large Language Models (LLMs) have revolutionized natural language processing tasks, but their deployment in datacenter environments faces significant challenges due to the massive memory requirements of key-value (KV) caches. During the autoregressive decoding process, KV caches consume substantial GPU memory, limiting...
CXL-SpecKV: A Disaggregated FPGA Speculative KV-Cache for Datacenter LLM Serving. Large Language Models (LLMs) have revolutionized natural language processing tasks, but their deployment in datacenter environments faces significant challenges due to the massive memory requirements of key-value (KV) caches. During the a...
ai_general
cs.AI
2025-12-11T15:40:36+00:00
2512.20034v1
VSA:Visual-Structural Alignment for UI-to-Code
The automation of user interface development has the potential to accelerate software delivery by mitigating intensive manual implementation. Despite the advancements in Large Multimodal Models for design-to-code translation, existing methodologies predominantly yield unstructured, flat codebases that lack compatibilit...
VSA:Visual-Structural Alignment for UI-to-Code. The automation of user interface development has the potential to accelerate software delivery by mitigating intensive manual implementation. Despite the advancements in Large Multimodal Models for design-to-code translation, existing methodologies predominantly yield uns...
ir_retrieval
cs.IR
2025-12-23T03:55:45+00:00
2512.22258v1
Logic Sketch Prompting (LSP): A Deterministic and Interpretable Prompting Method
Large language models (LLMs) excel at natural language reasoning but remain unreliable on tasks requiring strict rule adherence, determinism, and auditability. Logic Sketch Prompting (LSP) is a lightweight prompting framework that introduces typed variables, deterministic condition evaluators, and a rule based validato...
Logic Sketch Prompting (LSP): A Deterministic and Interpretable Prompting Method. Large language models (LLMs) excel at natural language reasoning but remain unreliable on tasks requiring strict rule adherence, determinism, and auditability. Logic Sketch Prompting (LSP) is a lightweight prompting framework that introdu...
ai_general
cs.AI
2025-12-24T09:20:35+00:00
2512.21512v1
Fixed-Threshold Evaluation of a Hybrid CNN-ViT for AI-Generated Image Detection Across Photos and Art
AI image generators create both photorealistic images and stylized art, necessitating robust detectors that maintain performance under common post-processing transformations (JPEG compression, blur, downscaling). Existing methods optimize single metrics without addressing deployment-critical factors such as operating p...
Fixed-Threshold Evaluation of a Hybrid CNN-ViT for AI-Generated Image Detection Across Photos and Art. AI image generators create both photorealistic images and stylized art, necessitating robust detectors that maintain performance under common post-processing transformations (JPEG compression, blur, downscaling). Exis...
cv_vision
cs.CV
2025-12-25T05:19:09+00:00
2512.23956v1
Implicit geometric regularization in flow matching via density weighted Stein operators
Flow Matching (FM) has emerged as a powerful paradigm for continuous normalizing flows, yet standard FM implicitly performs an unweighted $L^2$ regression over the entire ambient space. In high dimensions, this leads to a fundamental inefficiency: the vast majority of the integration domain consists of low-density ``vo...
Implicit geometric regularization in flow matching via density weighted Stein operators. Flow Matching (FM) has emerged as a powerful paradigm for continuous normalizing flows, yet standard FM implicitly performs an unweighted $L^2$ regression over the entire ambient space. In high dimensions, this leads to a fundament...
stat_ml
stat.ML
2025-12-30T03:08:00+00:00
2511.18530v1
Transforming Conditional Density Estimation Into a Single Nonparametric Regression Task
We propose a way of transforming the problem of conditional density estimation into a single nonparametric regression task via the introduction of auxiliary samples. This allows leveraging regression methods that work well in high dimensions, such as neural networks and decision trees. Our main theoretical result chara...
Transforming Conditional Density Estimation Into a Single Nonparametric Regression Task. We propose a way of transforming the problem of conditional density estimation into a single nonparametric regression task via the introduction of auxiliary samples. This allows leveraging regression methods that work well in high ...
stat_ml
stat.ML
2025-11-23T16:50:34+00:00
2512.22608v3
Beyond Isolated Investor: Predicting Startup Success via Roleplay-Based Collective Agents
Due to the high value and high failure rates of startups, predicting their success is a critical challenge. Existing approaches typically model startup success from a single decision-maker's perspective, overlooking the collective dynamics that dominate real-world venture capital (VC) decision-making. We propose SimVC-...
Beyond Isolated Investor: Predicting Startup Success via Roleplay-Based Collective Agents. Due to the high value and high failure rates of startups, predicting their success is a critical challenge. Existing approaches typically model startup success from a single decision-maker's perspective, overlooking the collectiv...
ai_general
cs.AI
2025-12-27T14:34:44+00:00
2510.16232v3
Personalized Collaborative Learning with Affinity-Based Variance Reduction
Multi-agent learning faces a fundamental tension: leveraging distributed collaboration without sacrificing the personalization needed for diverse agents. This tension intensifies when aiming for full personalization while adapting to unknown heterogeneity levels -- gaining collaborative speedup when agents are similar,...
Personalized Collaborative Learning with Affinity-Based Variance Reduction. Multi-agent learning faces a fundamental tension: leveraging distributed collaboration without sacrificing the personalization needed for diverse agents. This tension intensifies when aiming for full personalization while adapting to unknown he...
stat_ml
stat.ML
2025-10-17T21:49:51+00:00
2511.14763v3
Membership Inference Attack against Large Language Model-based Recommendation Systems: A New Distillation-based Paradigm
Membership Inference Attack (MIA) aims to determine whether a specific data sample was included in the training dataset of a target model. Traditional MIA approaches rely on shadow models to mimic target model behavior, but their effectiveness diminishes for Large Language Model (LLM)-based recommendation systems due t...
Membership Inference Attack against Large Language Model-based Recommendation Systems: A New Distillation-based Paradigm. Membership Inference Attack (MIA) aims to determine whether a specific data sample was included in the training dataset of a target model. Traditional MIA approaches rely on shadow models to mimic t...
ir_retrieval
cs.IR
2025-09-16T09:36:43+00:00
2512.06721v2
ProAgent: Harnessing On-Demand Sensory Contexts for Proactive LLM Agent Systems in the Wild
Recent studies have begun to explore proactive large language model (LLM) agents that provide unobtrusive assistance by automatically leveraging contextual information, such as in code editing and in-app suggestions. However, most focus on short, task-specific episodes or on-screen contexts, rather than continuously pe...
ProAgent: Harnessing On-Demand Sensory Contexts for Proactive LLM Agent Systems in the Wild. Recent studies have begun to explore proactive large language model (LLM) agents that provide unobtrusive assistance by automatically leveraging contextual information, such as in code editing and in-app suggestions. However, m...
ai_general
cs.AI
2025-12-07T08:21:07+00:00
2512.19088v1
Retrieving Objects from 3D Scenes with Box-Guided Open-Vocabulary Instance Segmentation
Locating and retrieving objects from scene-level point clouds is a challenging problem with broad applications in robotics and augmented reality. This task is commonly formulated as open-vocabulary 3D instance segmentation. Although recent methods demonstrate strong performance, they depend heavily on SAM and CLIP to g...
Retrieving Objects from 3D Scenes with Box-Guided Open-Vocabulary Instance Segmentation. Locating and retrieving objects from scene-level point clouds is a challenging problem with broad applications in robotics and augmented reality. This task is commonly formulated as open-vocabulary 3D instance segmentation. Althoug...
cv_vision
cs.CV
2025-12-22T06:57:42+00:00