KOINEU

No Free Lunch in Language Model Bias Mitigation? Targeted Bias Reduction Can Exacerbate Unmitigated LLM Biases

본 논문은 대형 언어 모델(LLM)의 편향 완화가 단일 차원에서의 성공에 머무르지 않고, 다른 차원에서 새로운 편향을 유발하거나 기존 편향을 심화시킬 수 있다는 중요한 교차 효과를 체계적으로 탐구한다. 연구진은 먼저 7개의 서로 다른 모델 패밀리(예: GPT, BERT, T5 등)에서 파생된 10개의 모델을 선정하고, 각각에 대해 네 가지 대표적인 편향 완화 기법(데이터 재샘플링, 손실 가중치 조정, 사후 필터링, 프롬프트 엔지니어링)을 적용하였다. 이때 인종, 종교, 직업·성별이라는 세 가지 주요 편향 축을 설정하고, 각 축에 대

No Free Lunch in Language Model Bias Mitigation? Targeted Bias Reduction Can Exacerbate Unmitigated LLM Biases

NystagmusNet: Explainable Deep Learning for Photosensitivity Risk Prediction

ShareChat: A Dataset of Chatbot Conversations in the Wild

Thucy: An LLM-based Multi-Agent System for Claim Verification across Relational Databases

Visual Sync: Multi-Camera Synchronization via Cross-View Object Motion

Wearable-informed generative digital avatars predict task-conditioned post-stroke locomotion

When AI Bends Metal: AI-Assisted Optimization of Design Parameters in Sheet Metal Forming

Domain-Specific Foundation Model Improves AI-Based Analysis of Neuropathology

NVIDIA Nemotron 3: Efficient and Open Intelligence

Connectivity-Preserving Cortical Surface Tetrahedralization

Convergence of Outputs When Two Large Language Models Interact in a Multi-Agentic Setup

Planning as Descent: Goal-Conditioned Latent Trajectory Synthesis in Learned Energy Landscapes

World Models for Autonomous Navigation of Terrestrial Robots from LIDAR Observations

Deep Research: A Systematic Survey

Proportional integral derivative booster for neural networks-based time-series prediction: Case of water demand prediction

Satisfiability Modulo Theory Meets Inductive Logic Programming

Catching UX Flaws in Code: Leveraging LLMs to Identify Usability Flaws at the Development Stage

Efficient Kernel Mapping and Comprehensive System Evaluation of LLM Acceleration on a CGLA

A gradient descent algorithm for computing circle patterns

CRAFT-E: A Neuro-Symbolic Framework for Embodied Affordance Grounding

ELANA: A Simple Energy and Latency Analyzer for LLMs

Evolutionary Architecture Search through Grammar-Based Sequence Alignment

FIN-bench-v2: A Unified and Robust Benchmark Suite for Evaluating Finnish Large Language Models

From Theory of Mind to Theory of Environment: Counterfactual Simulation of Latent Environmental Dynamics

Improving VQA Reliability: A Dual-Assessment Approach with Self-Reflection and Cross-Model Verification

Suzume-chan: Your Personal Navigator as an Embodied Information Hub

The body is not there to compute: Comment on 'Informational embodiment: Computational role of information structure in codes and robots' by Pitti et al

The Erosion of LLM Signatures: Can We Still Distinguish Human and LLM-Generated Scientific Ideas After Iterative Paraphrasing?

The stationary focus of the Kiepert parabola over a special Poncelet triangle family

A Linear Expectation Constraint for Selective Prediction and Routing with False-Discovery Control

An AI Monkey Gets Grapes for Sure -- Sphere Neural Networks for Reliable Decision-Making

AncientBench: Towards Comprehensive Evaluation on Excavated and Transmitted Chinese Corpora

Arc Spline Approximation of Envelopes of Evolving Planar Domains

Beyond Additivity: Sparse Isotonic Shapley Regression toward Nonlinear Explainability

Dynamically Scaled Activation Steering

E3AD: An Emotion-Aware Vision-Language-Action Model for Human-Centric End-to-End Autonomous Driving

ElecTwit: A Framework for Studying Persuasion in Multi-Agent Social Systems

Enhancing Cross Domain SAR Oil Spill Segmentation via Morphological Region Perturbation and Synthetic Label-to-SAR Generation

Evolving CNN Architectures: From Custom Designs to Deep Residual Models for Diverse Image Classification and Detection Tasks

From Compound Figures to Composite Understanding: Developing a Multi-Modal LLM from Biomedical Literature with Medical Multiple-Image Benchmarking and Validation

From Isolation to Entanglement: When Do Interpretability Methods Identify and Disentangle Known Concepts?

GRC-Net: Gram Residual Co-attention Net for epilepsy prediction

Gutenberg-Richter-like relations in physical systems

HealthContradict: Evaluating Biomedical Knowledge Conflicts in Language Models

Hybrid Stackelberg Game and Diffusion-based Auction for Two-tier Agentic AI Task Offloading in Internet of Agents

Interpretable Plant Leaf Disease Detection Using Attention-Enhanced CNN

Lang3D-XL: Language Embedded 3D Gaussians for Large-scale Scenes

Length-Aware Adversarial Training for Variable-Length Trajectories: Digital Twins for Mall Shopper Paths

Log Anomaly Detection with Large Language Models via Knowledge-Enriched Fusion

MAR-FL: A Communication Efficient Peer-to-Peer Federated Learning System

< Category Statistics (Total: 5502) >

Start searching

No results found