Abstract: Machine Unlearning is an upcoming field, with tremendous potential for practical impact. While the theoretically rigorous algorithms are still preliminary, there has been a plethora of empirical works providing various unlearning algorithms for various deep-learning settings. We revisit the efficacy of several practical methods for approximate machine unlearning developed for large-scale deep learning. We experimentally demonstrate that, while existing unlearning methods have been demonstrated to be effective in a number of evaluation settings (e.g., alleviating membership inference attacks), they fail to remove the effects of data poisoning, across a variety of types of poisoning attacks (indiscriminate, targeted, and a newly-introduced Gaussian poisoning attack) and models (image classifiers And LLMs); even when granted a relatively large compute budget. In order to precisely characterize unlearning efficacy, we introduce new evaluation metrics for unlearning based on data poisoning. Our results suggest that a broader perspective, including a wider variety of evaluations, is required to avoid a false sense of confidence in machine unlearning procedures for deep learning without provable guarantees. Moreover, while unlearning methods show some signs of being useful to efficiently remove poisoned data points without having to retrain, our work suggests that these methods are not yet "ready for prime time" and currently provide limited benefit over retraining. 
The talk will be based on this paper - Current Machine Unlearning Fails to Remove Data Poisoning Attack

Brief Bio:  Ayushi s a postdoc researcher working with Prof. Sasha Rakhlin at MIT. He completed his Ph.D. from the Computer Science department at Cornell University, advised by Professor Karthik Sridharan and Professor Robert D. Kleinberg. His research interests span stochastic optimization, reinforcement learning, and machine unlearning. Before his Ph.D., he spent a year at Google as a part of the inaugural Brain residency program. Before Google, he completed his undergraduate studies in computer science at IIT Kanpur in Indiawhere he was awarded the President's Gold medal. Ayush has also been a winner of a student best paper award at COLT 2019. Ayush is on the academic job market for the 2024/25cycle.