Four papers in ICLR’24: (1) Machine unlearning for safe image generation; (2) DeepZero: Training neural networks from scratch using only forward passes; (3) Backdoor data sifting; (4) Visual prompting automation
OPTML Lab’s latest tutorial, accepted for AAAI-24, spotlights the emerging field of zeroth-order machine learning (ZO-ML). Titled “Zeroth-Order Machine Learning: Fundamental Principles and Emerging Applications in Foundation Models,” this session will explore ZO-ML’s core theories and methodologies. It aims to showcase the integration of ZO-ML in advanced AI applications, particularly within foundation models, highlighting its potential to revolutionize current AI paradigms. This tutorial represents a significant step in marrying theoretical insights with practical AI solutions.
