newsA research position paper examines the integration of Large Language Models (LLMs) in agent-based social simulations, highlighting both opportunities and limitations. The study proposes Hybrid Constitutional Architectures that combine classical agent-based models with small language models and LLMs to balance expressive flexibility with analytical transparency.
Researchers introduced VAF, a systematic evaluation pipeline to measure how visual web elements influence AI agent decision-making. The study tested 48 variants across 5 real-world websites and found that background contrast, item size, position, and card clarity significantly impact agent behavior, while font styling and text color have minimal effects.
Researchers have introduced fEDM+, an enhanced fuzzy ethical decision-making framework for AI systems that provides principle-level explainability and validates decisions against multiple stakeholder perspectives. The framework extends the original fEDM by adding transparent explanations of ethical decisions and replacing single-point validation with pluralistic validation that accommodates different ethical viewpoints.
Researchers developed M3TR, a new AI framework that uses temporal retrieval and multi-modal analysis to predict micro-video popularity with 19.3% better accuracy than existing methods. The system combines a Mamba-Hawkes Process module to model user feedback patterns with temporal-aware retrieval to identify historically relevant videos based on content and popularity trajectories.
Researchers analyzed user misconceptions about LLM-based programming assistants like ChatGPT, finding users often have misplaced expectations about web access, code execution, and debugging capabilities. The study examined Python programming conversations from WildChat dataset and identified the need for clearer communication of tool capabilities to prevent over-reliance and unproductive practices.
Researchers developed LEC-KG, a new framework that combines Large Language Models with Knowledge Graph Embeddings to better extract and structure information from unstructured text. The system was tested on Chinese Sustainable Development Goal reports and showed significant improvements over traditional LLM approaches, particularly for identifying rare relationships in domain-specific content.
Researchers developed RL-CMSA, a hybrid reinforcement learning approach for solving the min-max Multiple Traveling Salesman Problem that combines probabilistic clustering, exact optimization, and solution refinement. The method outperforms existing algorithms by balancing exploration and exploitation to minimize the longest tour across multiple salesmen.
Researchers developed QD-MAPPER, a framework using Quality Diversity algorithms and Neural Cellular Automata to automatically generate diverse maps for evaluating Multi-Agent Path Finding (MAPF) algorithms. This addresses the limitation of testing MAPF algorithms on fixed, human-designed maps that may not cover all scenarios and could lead to overfitting.
Researchers introduce AudioCapBench, a new benchmark for evaluating how well large multimodal AI models can generate captions for audio content across sound, music, and speech domains. The study tested 13 models from OpenAI and Google Gemini, finding that Gemini models generally outperformed OpenAI in overall captioning quality, though all models struggled most with music captioning.
Researchers propose BDGxRL, a novel framework using Diffusion Schrödinger Bridge to enable reinforcement learning agents to transfer policies across different domains without direct target environment access. The method aligns source domain transitions with target dynamics through offline demonstrations and introduces reward modulation for consistent learning.
Researchers propose the Intrinsic Lorentz Neural Network (ILNN), a fully intrinsic hyperbolic architecture that performs all computations within the Lorentz model for better handling of hierarchical data structures. The network introduces novel components including point-to-hyperplane layers and GyroLBN batch normalization, achieving state-of-the-art performance on CIFAR and genomic benchmarks while outperforming Euclidean baselines.
Researchers introduce ARGUS, a framework for studying how narrative features influence persuasion in online arguments. The study analyzes a ChangeMyView corpus using both traditional classifiers and large language models to identify which storytelling elements make arguments more convincing.
Researchers introduce DirMixE, a new machine learning approach for handling test-agnostic long-tail recognition problems where test data distributions are unknown and imbalanced. The method uses a hierarchical Mixture-of-Expert strategy with Dirichlet meta-distributions and includes a Latent Skill Finetuning framework for efficient parameter tuning of foundation models.
Researchers propose LLM-hRIC, a new framework that combines large language models with hierarchical radio access network intelligent controllers to improve O-RAN networks. The system uses LLM-powered non-real-time controllers for strategic guidance and reinforcement learning for near-real-time decision making in network management.
Researchers conducted interviews with 11 practitioners at major tech companies to study how fairness considerations are integrated into recommender system workflows. The study identified key challenges including defining fairness in RS contexts, balancing stakeholder interests, and facilitating cross-team communication between technical, legal, and fairness teams.
Researchers introduce iterated Shared Q-Learning (iS-QL), a new reinforcement learning method that bridges target-free and target-based approaches by using only the last linear layer as a target network while sharing other parameters. The technique achieves comparable performance to traditional target-based methods while maintaining the memory efficiency of target-free approaches.
Researchers introduce pact, a new SMT model counter that can handle hybrid formulas containing both discrete and continuous variables using hashing-based approximate counting. The tool significantly outperforms existing baselines, successfully processing 456 out of 3119 test instances compared to only 83 for the baseline method.
Researchers introduce FLOP, a new causal discovery algorithm for linear models that significantly reduces computation time through fast parent selection and Cholesky-based score updates. The algorithm achieves near-perfect accuracy in standard benchmarks and makes discrete search approaches viable for causal structure learning.
Researchers analyzed DINOv2 vision transformer using Sparse Autoencoders to understand how it processes visual information, discovering that the model uses specialized concept dictionaries for different tasks like classification and segmentation. They propose the Minkowski Representation Hypothesis as a new framework for understanding how vision transformers combine conceptual archetypes to form representations.
Researchers introduce USplat4D, a new uncertainty-aware dynamic Gaussian Splatting framework that improves 3D scene reconstruction from monocular video by modeling per-Gaussian uncertainty. The approach addresses motion drift and poor synthesis quality by treating well-observed Gaussians as reliable anchors while handling poorly observed ones as less reliable.
Researchers have developed MEDIC, a neural network framework for Data Quality Monitoring (DQM) in particle physics experiments that uses machine learning to automatically detect detector anomalies and identify malfunctioning components. The simulation-driven approach using modified Delphes detector simulation represents an initial step toward comprehensive ML-based DQM systems for future particle detectors.
Researchers propose an enhanced methodology using rough set theory to improve explainability of Graph Spectral Clustering (GSC) algorithms. The approach addresses challenges in explaining clustering results, particularly when applied to text documents where spectral space embeddings lack clear relation to content.
Researchers introduce CSyMR-Bench, a new benchmark for evaluating AI systems' ability to perform complex music information retrieval tasks from symbolic notation. The benchmark includes 126 multiple-choice questions requiring compositional reasoning, and demonstrates that tool-augmented AI approaches outperform language model-only methods by 5-7%.
Researchers studied how personality-trait-infused LLM messaging affects user perceptions in behavior change systems. The study found that personality-based personalization works through aggregate exposure patterns rather than individual message optimization, with users rating personality-informed messages as more personalized and appropriate.
Researchers introduce FedVG, a new federated learning framework that uses gradient-guided aggregation and global validation sets to improve model performance in distributed training environments. The approach addresses client drift issues in heterogeneous data settings and can be integrated with existing federated learning algorithms.