🏢 Korea Advanced Institute of Science and Technology
ReMoDetect: Reward Models Recognize Aligned LLM's Generations
·4799 words·23 mins·
loading
·
loading
AI Generated
Natural Language Processing
Large Language Models
🏢 Korea Advanced Institute of Science and Technology
ReMoDetect leverages reward models to identify and classify LLM-generated text. By using continual preference fine-tuning and incorporating human/LLM mixed text, ReMoDetect achieves state-of-the-art p…
Pretraining with Random Noise for Fast and Robust Learning without Weight Transport
·1659 words·8 mins·
loading
·
loading
Machine Learning
Deep Learning
🏢 Korea Advanced Institute of Science and Technology
Random noise pretraining dramatically speeds up and enhances neural network learning without weight transport, mimicking the brain’s developmental process and achieving performance comparable to backp…
MV2Cyl: Reconstructing 3D Extrusion Cylinders from Multi-View Images
·3293 words·16 mins·
loading
·
loading
Computer Vision
3D Vision
🏢 Korea Advanced Institute of Science and Technology
MV2Cyl: A novel method reconstructs 3D extrusion cylinder CAD models directly from multi-view images, surpassing accuracy of methods using raw 3D geometry.
Genetic-guided GFlowNets for Sample Efficient Molecular Optimization
·3431 words·17 mins·
loading
·
loading
Machine Learning
Deep Learning
🏢 Korea Advanced Institute of Science and Technology
Genetic-guided GFlowNets revolutionize sample-efficient molecular optimization by smartly integrating genetic algorithms into GFlowNets training, achieving state-of-the-art performance with substantia…
A Simple Remedy for Dataset Bias via Self-Influence: A Mislabeled Sample Perspective
·3350 words·16 mins·
loading
·
loading
AI Theory
Fairness
🏢 Korea Advanced Institute of Science and Technology
This paper introduces Bias-Conditioned Self-Influence (BCSI) for precise bias-conflicting sample detection and model rectification, enhancing fairness in machine learning.
A Kernel Perspective on Distillation-based Collaborative Learning
·2168 words·11 mins·
loading
·
loading
Machine Learning
Federated Learning
🏢 Korea Advanced Institute of Science and Technology
This paper introduces DCL-KR and DCL-NN, novel distillation-based collaborative learning algorithms achieving nearly minimax optimal convergence rates in heterogeneous environments without direct data…