Machine unlearning -- efficiently removing the effect of a small "forget set" of training data on a pre-trained machine learning model -- has recently attracted significant research interest. Despite this interest, however, recent work shows that existing machine unlearning techniques do not hold up to thorough evaluation in non-convex settings. In this work, we introduce a new machine unlearning technique that exhibits strong empirical performance even in such challenging settings. Our starting point is the perspective that the goal of unlearning is to produce a model whose outputs are statistically indistinguishable from those of a model re-trained on all but the forget set. This perspective naturally suggests a reduction from the unlearning problem to that of data attribution, where the goal is to predict the effect of changing the training set on a model's outputs. Thus motivated, we propose the following meta-algorithm, which we call Datamodel Matching (DMM): given a trained model, we (a) use data attribution to predict the output of the model if it were re-trained on all but the forget set points; then (b) fine-tune the pre-trained model to match these predicted outputs. In a simple convex setting, we show how this approach provably outperforms a variety of iterative unlearning algorithms. Empirically, we use a combination of existing evaluations and a new metric based on the KL-divergence to show that even in non-convex settings, DMM achieves strong unlearning performance relative to existing algorithms. An added benefit of DMM is that it is a meta-algorithm, in the sense that future advances in data attribution translate directly into better unlearning algorithms, pointing to a clear direction for future progress in unlearning.
BIFOLD & TU Berlin(BIFOLD与柏林工业大学);CISPA Helmholtz Center for Information Security(CISPA亥姆霍兹信息安全中心);University of Illinois Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校);Microsoft Research(微软研究院);The University of Melbourne(墨尔本大学);University of Toronto & Vector Institute(多伦多大学与向量研究所)
Machine Unlearning: Taxonomy, Metrics, Applications, Challenges, and Prospects
Nanjing University of Science and Technology(南京理工大学);Xidian University(西安电子科技大学);CSIRO(联邦科学与工业研究组织);University of Western Australia(西澳大学);University of Technology Sydney(悉尼科技大学)
Towards Federated Domain Unlearning: Verification Methodologies and Challenges
University of Macau(澳门大学);University Of Electronic Science And Technology(电子科技大学);Institute of High Performance Computing, Agency for Science, Technology and Research(新加坡科技研究局高性能计算研究院)
Certified Unlearning in Decentralized Federated Learning
School of Computer Science and Technology, Shandong University, Qingdao, China(山东大学计算机科学与技术学院);School of Electrical Engineering and Computer Science, TU Berlin, Berlin, Germany(柏林技术大学电气工程与计算机科学学院);College of Computer Science and Technology, Ocean University of China, Qingdao, China(中国海洋大学计算机科学与技术学院)
Causal Unlearning in Collaborative Optimization: Exact and Approximate Influence Reversal under Adversarial Contributions
Department of Computer Engineering, SRC, Islamic Azad University Tehran, Iran(伊朗伊斯兰Azad大学塔希尔分校计算机工程系);School of Computer Engineering, Iran University of Science and Technology Tehran, Iran(伊朗科学技术大学塔希尔分校计算机工程系);Meta CA, USA(美国Meta公司)
King Abdullah University of Science and Technology(阿卜杜拉国王科技大学);SDAIA-KAUST Center of Excellence in Data Science and Artificial Intelligence(沙特数据与人工智能局-阿卜杜拉国王科技大学数据科学与人工智能卓越中心);Iowa State University(爱荷华州立大学);Computational Bioscience Research Center(计算生物科学研究中心)
DeepObliviate: A Powerful Charm for Erasing Data Residual Memory in Deep Neural Networks
SKLOIS, Institute of Information Engineering, Chinese Academy of Sciences(中国科学院信息工程研究所智能信息处理实验室);School of Cyber Security, University of Chinese Academy of Sciences(中国科学院大学网络空间安全学院)
Descend or Rewind? Stochastic Gradient Descent Unlearning
Department of Engineering Sciences and Applied Mathematics, Northwestern University, Evanston, IL, United States(工程科学与应用数学系,西北大学,埃文斯顿,伊利诺伊州,美国);Department of Industrial Engineering(工业工程系);Applied Mathematics, Northwestern University, Evanston, IL, United States(应用数学,西北大学,埃文斯顿,伊利诺伊州,美国)
The Chinese University of Hong Kong, Shenzhen(香港中文大学(深圳));Zhejiang University(浙江大学);University of Sydney(悉尼大学);Northwestern University(西北大学);Analogy AI, Inc.(Analogy AI公司);Mohamed Bin Zayed University of Artificial Intelligence(穆罕默德·本·扎耶德人工智能大学)
Forgetting to Witness: Efficient Federated Unlearning and Its Visible Evaluation
School of Cyber Security, University of Chinese Academy of Sciences(中国科学院大学网络空间安全学院);Institute of Information Engineering, Chinese Academy of Sciences(中国科学院信息工程研究所);King Abdullah University of Science and Technology(阿卜杜拉国王科技大学)
Enabling Group Fairness in Graph Unlearning via Bi-level Debiasing
Department of Electrical Engineering and Computer Science, University of California, Irvine(电气工程与计算机科学系,加州大学伊尔凡分校);Donald Bren School of Information and Computer Sciences, University of California, Irvine(戴维·布伦信息与计算机科学学院,加州大学伊尔凡分校)