Machine Unlearning via Simulated Oracle Matching

Despite increasing interest in machine unlearning, recent work shows that under strong evaluations, existing techniques largely fail to unlearn in non-convex settings. In this paper, we introduce a new technique for machine unlearning in such settings. Key to our method is a reduction from the problem of machine unlearning to that of data attribution. In particular, we show theoretically (in an underdetermined regression setting) and empirically (in a standard deep learning setting) that given access to the outputs of a perfectly unlearned model (i.e., a model trained from scratch on the non-unlearned data), we can quickly fine-tune an existing model on these predictions and match the target model predictions out-of-sample. Meanwhile, predicting such oracle'' outputs is precisely the goal of a recent line of work in data attribution called {\em datamodeling}. Combining these two insights yields an end-to-end unlearning algorithm in which one first {\em predicts the output} of a model re-trained from scratch, then {\em fine-tunes} an existing model to match these predicted outputs. Across different types and sizes of forget sets, we show that this two-stage algorithm results in strong unlearning performance being close to indistinguishable from the fully-retrainedoracle'' model in some cases. As an added benefit, our reduction means that future improvements to data attribution---whether in accuracy or efficiency---may in turn yield better unlearning algorithms.