Improved sample complexity bounds for agnostic binary classification in the tau-based model!

This paper presents CoUn, a contrastive learning (CL)-based machine unlearning (MU) framework using only retain data. Further, our proposed CL module can be integrated with existing baselines to empower their performance.

A pathbreaking model based on neural differential equations with rigorous mathematical foundation

This paper proposes the ACT data pipeline, which reduces human annotation costs by using MLLMs as annotators and error detectors, and provides a theoretical analysis to ensure effective downstream training.