KOINEU

Differences That Matter: Auditing Models for Capability Gap Discovery and Rectification

이 논문은 다모달 대형 언어 모델(MLLM)의 평가와 개선을 위한 새로운 접근법을 제시하고 있습니다. 기존의 MLLM 평가 방법들은 해석성이 부족하여 모델 간의 능력 격차를 명확하게 드러내지 못하는 문제점이 있었습니다. 이를 해결하기 위해 연구자들은 AuditDM이라는 자동화된 프레임워크를 개발하였습니다. AuditDM은 강화학습을 통해 MLLM을 감사원으로 미세 조정하여, 모델 간의 불일치를 최대한 증가시키는 어려운 질문과 반사실적 이미지를 생성합니다. 이를 통해 다양한 실패 유형을 찾아내고, 이러한 실패 사례들은 주석 없이도 데

Differences That Matter: Auditing Models for Capability Gap Discovery and Rectification

Directional Optimization Asymmetry in Transformers: A Synthetic Stress Test

Dominating vs. Dominated: Generative Collapse in Diffusion Models

Dual Attention Guided Defense Against Malicious Edits

EmeraldMind: A Knowledge Graph-Augmented Framework for Greenwashing Detection

ESPADA: Execution Speedup via Semantics Aware Demonstration Data Downsampling for Imitation Learning

From In Silico to In Vitro: Evaluating Molecule Generative Models for Hit Generation

Generative Adversarial Reasoner: Enhancing LLM Reasoning with Adversarial Reinforcement Learning

Graph-Based Bayesian Optimization for Quantum Circuit Architecture Search with Uncertainty Calibrated Surrogates

Hardware Software Optimizations for Fast Model Recovery on Reconfigurable Architectures

hls4ml: A Flexible, Open-Source Platform for Deep Learning Acceleration on Reconfigurable Hardware

Image Complexity-Aware Adaptive Retrieval for Efficient Vision-Language Models

Kardia-R1: Unleashing LLMs to Reason toward Understanding and Empathy for Emotional Support via Rubric-as-Judge Reinforcement Learning

Large Language Models as Pokémon Battle Agents: Strategic Play and Content Generation

LLM-Driven Feature-Level Adversarial Attacks on Android Malware Detectors

Mosaic Pruning: A Hierarchical Framework for Generalizable Pruning of Mixture-of-Experts Models

Multi-agent Adaptive Mechanism Design

Multi-Rigid-Body Approximation of Human Hands with Application to Digital Twin

NeuralRemaster: Phase-Preserving Diffusion for Structure-Aligned Generation

Optimal Software Pipelining and Warp Specialization for Tensor Core GPUs

Pix2NPHM: Learning to Regress NPHM Reconstructions From a Single Image

PULSE: A Unified Multi-Task Architecture for Cardiac Segmentation, Diagnosis, and Few-Shot Cross-Modality Clinical Adaptation

Quantifying Uncertainty in Machine Learning-Based Pervasive Systems: Application to Human Activity Recognition

Refusal Steering: Fine-grained Control over LLM Refusal Behaviour for Sensitive Topics

Relaxed Conditions for Secrecy in a Role-Based Specification

SALVE: Sparse Autoencoder-Latent Vector Editing for Mechanistic Control of Neural Networks

Seismology modeling agent: A smart assistant for geophysical researchers

SpidR-Adapt: A Universal Speech Representation Model for Few-Shot Adaptation

Step-Tagging: Toward controlling the generation of Language Reasoning Models through step monitoring

Strategic Self-Improvement for Competitive Agents in AI Labour Markets

The Dead Salmons of AI Interpretability

Thermal RGB Fusion for Micro-UAV Wildfire Perimeter Tracking with Minimal Comms

ThinkTrap: Denial-of-Service Attacks against Black-box LLM Services via Infinite Thinking

TrafficSimAgent: A Hierarchical Agent Framework for Autonomous Traffic Simulation with MCP Control

Unified Interactive Multimodal Moment Retrieval via Cascaded Embedding-Reranking and Temporal-Aware Score Fusion

Universal Hirschberg for Width Bounded Dynamic Programs

VLM in a flash: I/O-Efficient Sparsification of Vision-Language Model via Neuron Chunking

When to compute in space

Measuring What Matters: Scenario-Driven Evaluation for Trajectory Predictors in Autonomous Driving

State Space Models for Bioacoustics: A comparative Evaluation with Transformers

TrajMoE: Scene-Adaptive Trajectory Planning with Mixture of Experts and Reinforcement Learning

Large Language Models as Discounted Bayesian Filters

Mitigating Gender Bias in Depression Detection via Counterfactual Inference

Enabling Conversational Behavior Reasoning Capabilities in Full-Duplex Speech

iOS as Acceleration

Mechanistic Interpretability of GPT-2: Lexical and Contextual Layers in Sentiment Analysis

Neural Networks for Predicting Permeability Tensors of 2D Porous Media: Comparison of Convolution- and Transformer-based Architectures

MindFuse: Towards GenAI Explainability in Marketing Strategy Co-Creation

Embodied Co-Design for Rapidly Evolving Agents: Taxonomy, Frontiers, and Challenges

Evidence-Driven Decision Support for AI Model Selection in Research Software Engineering

< Category Statistics (Total: 2829) >

Start searching

No results found