Vision-Language Models are Strong Noisy Label Detectorsby Tong Wei, Hao-Tian Li, Chun-Shu Li, Jiang-Xin Shi,…
Unified Gradient-Based Machine Unlearning with Remain Geometry Enhancementby Zhehao Huang, Xinwen Cheng, JingHao Zheng, Haoran…
Balancing Cost and Effectiveness of Synthetic Data Generation Strategies for LLMsby Yung-Chieh Chan, George Pu,…
Scalable Fine-tuning from Multiple Data Sources: A First-Order Approximation Approachby Dongyue Li, Ziniu Zhang, Lu…
On the Inductive Bias of Stacking Towards Improving Reasoningby Nikunj Saunshi, Stefani Karp, Shankar Krishnan,…
Localizing Memorization in SSL Vision Encodersby Wenhao Wang, Adam Dziedzic, Michael Backes, Franziska BoenischFirst submitted…
DiaSynth: Synthetic Dialogue Generation Framework for Low Resource Dialogue Applicationsby Sathya Krishnan Suresh, Wu Mengjun,…
Understanding the Benefits of SimCLR Pre-Training in Two-Layer Convolutional Neural Networksby Han Zhang, Yuan CaoFirst…
DisGeM: Distractor Generation for Multiple Choice Questions with Span Maskingby Devrim Cavusoglu, Secil Sen, Ulas…
SOAR: Self-supervision Optimized UAV Action Recognition with Efficient Object-Aware Pretrainingby Ruiqi Xian, Xiyang Wu, Tianrui…