AI
-
Explicit v.s. Implicit Memory: Exploring Multi-hop Complex Reasoning Over Personalized Information
Source: arXiv AI PapersRead more: Explicit v.s. Implicit Memory: Exploring Multi-hop Complex Reasoning Over Personalized InformationThis paper introduces a multi-hop personalized reasoning task to evaluate memory mechanisms in large language model agents for complex reasoning…
-
“DIVE” into Hydrogen Storage Materials Discovery with AI Agents
Source: arXiv AI PapersRead more: “DIVE” into Hydrogen Storage Materials Discovery with AI AgentsThe DIVE multi-agent AI workflow enhances the extraction of experimental data from scientific literature, significantly improving the discovery of solid-state…
-
SPANER: Shared Prompt Aligner for Multimodal Semantic Representation
Source: arXiv AI PapersRead more: SPANER: Shared Prompt Aligner for Multimodal Semantic RepresentationSPANER is a new modality-agnostic framework that aligns multimodal embeddings into a unified semantic space using a shared prompt mechanism.…
-
Toward Better EHR Reasoning in LLMs: Reinforcement Learning with Expert Attention Guidance
Source: arXiv AI PapersRead more: Toward Better EHR Reasoning in LLMs: Reinforcement Learning with Expert Attention GuidanceThe paper introduces EAG-RL, a novel two-stage training framework that enhances large language models’ (LLMs) reasoning abilities on electronic health…
-
Breaking the SFT Plateau: Multimodal Structured Reinforcement Learning for Chart-to-Code Generation
Source: arXiv AI PapersRead more: Breaking the SFT Plateau: Multimodal Structured Reinforcement Learning for Chart-to-Code GenerationThis paper addresses the limitations of supervised fine-tuning (SFT) in chart-to-code generation by introducing Multimodal Structured Reinforcement Learning (MSRL). MSRL…
-
V2P: From Background Suppression to Center Peaking for Robust GUI Grounding Task
Source: arXiv AI PapersRead more: V2P: From Background Suppression to Center Peaking for Robust GUI Grounding TaskThe V2P method improves GUI element localization by addressing background distractions and center-edge distinction issues in attention mechanisms. This approach…
-
Interactive Query Answering on Knowledge Graphs with Soft Entity Constraints
Source: arXiv AI PapersRead more: Interactive Query Answering on Knowledge Graphs with Soft Entity ConstraintsThe paper introduces a novel approach for query answering on incomplete knowledge graphs by incorporating soft entity constraints, addressing the…
-
ITL-LIME: Instance-Based Transfer Learning for Enhancing Local Explanations in Low-Resource Data Settings
Source: arXiv AI PapersRead more: ITL-LIME: Instance-Based Transfer Learning for Enhancing Local Explanations in Low-Resource Data SettingsThe paper introduces ITL-LIME, an instance-based transfer learning framework designed to improve the fidelity and stability of local explanations in…
-
Preliminary suggestions for rigorous GPAI model evaluations
Source: arXiv AI PapersRead more: Preliminary suggestions for rigorous GPAI model evaluationsThis document offers preliminary suggestions for rigorous evaluation practices of general-purpose AI (GPAI) models to enhance validity and reproducibility. It…
-
Utilizing the RAIN method and Graph SAGE Model to Identify Effective Drug Combinations for Gastric Neoplasm Treatment
Source: arXiv AI PapersRead more: Utilizing the RAIN method and Graph SAGE Model to Identify Effective Drug Combinations for Gastric Neoplasm TreatmentThe study utilizes the RAIN method combined with the Graph SAGE model to identify effective drug combinations for treating gastric…