Approximate Data Deletion from Machine Learning Models: Algorithms and Evaluations

By Zachary Izzo, Mary Anne Smart, Kamalika Chaudhuri, James Y. Zou. 2020

379 citations recorded by the source corpus.

Deleting data from a trained machine learning (ML) model is a critical task in many applications. For example, we may want to remove the influence of training points that might be out of date or outliers. Regulations such as EU's General Data Protection Regulation also stipulate that individuals can request to have their data deleted. The naive approach to data deletion is to retrain the ML model on the remaining data, but this is too time consuming. Moreover there is no known efficient algorithm that exactly deletes data from most ML models. In this work, we evaluate several approaches for approximate data deletion from trained models. For the case of linear regression, we propose a new method with linear dependence on the feature dimension $d$, a significant gain over all existing methods which all have superlinear time dependence on the dimension. We also provide a new test for evaluating data deletion from linear models.

Subfields: Artificial Intelligence

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Privacy in Deep Learning: A Survey (2020, background): “hip inference attacks to the generalization gap, the overﬁtting of the model, and data memorization capabilities of neural networks. Deep neural networks have been shown to memorize the training data [53, 54, 55], rather than learning the latent properties of the data, which means they often tend to”
Mixed-Privacy Forgetting in Deep Networks (2020, methodology): “4 [24] Zachary Izzo, Mary Anne Smart, Kamalika Chaudhuri, and James Zou.”
Measuring Forgetting of Memorized Training Examples (2022, neutral): “Algorithms for machine unlearning have been proposed for k-means [24], empirical risk minimization [29, 35, 59, 76, 67], and deep learning [18, 26, 25, 60, 5].”
A Survey of Machine Unlearning (2022, neutral): “They approximated a Hessian matrix computation with a project residual update [15, 73] that combines gradient methods with synthetic data.”
Descent-to-Delete: Gradient-Based Methods for Machine Unlearning (2020, methodology): “Our main result leverages a distributed optimization algorithm that partitions the data, separately optimizes on each partition, and then averages the parameters, analyzed by Zhang et al. (2012). Optimizing separately on different partitions of the data, and then aggregating the results is also a we”
DART: Data Addition and Removal Trees (2020, neutral): “More details on the data addition results can be found in Section B.4 of the appendix.”
Large Language Model Unlearning (2023, neutral): “…for classification models, including data-reversed training [Tarun et al., 2023, Liu et al., 2022, Chundawat et al., 2023], optimization-based unlearning [Guo et al., 2019, Neel et al., 2021] and influence function based approaches [Jia et al., 2023, Warnecke et al., 2021, Izzo et al., 2021].”
Detecting Pretraining Data from Large Language Models (2023, background): “…has emerged as a solution for purging data from machine learning models, and various machine unlearning methods have been introduced (Ginart et al., 2019; Liu et al., 2020; Wu et al., 2020; Bourtoule et al., 2021; Izzo et al., 2021; Sekhari et al., 2021; Gupta et al., 2021; Ye et al., 2022).”
Forgetting Outside the Box: Scrubbing Deep Networks of Information Accessible from Input-Output Observations (2020, methodology): “[22] provides a projective residual update based method using synthetic data points to delete data points for linear regression based models.”
SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation (2023, methodology): “For example, influence unlearning (Izzo et al., 2021; Warnecke et al., 2021), built upon the influence function analysis of training data points (Koh & Liang, 2017), exhibits high-performance variance due to the selection of hyperparameters required for influence function approximations, as well as…”
Remember What You Want to Forget: Algorithms for Machine Unlearning (2021, methodology): “This relaxation has inspired the design of several eﬃcient algorithms for data deletion from machine learning models (Guo et al., 2020; Izzo et al., 2021; Neel et al., 2021; Ullah et al., 2021).”
Advances, challenges and opportunities in creating data for trustworthy AI (2022, neutral): “Similarly, data erasure can be challenging to implement, as large AI models memorize user information 106 , and selective erasure can magnify coverage gaps in the data 107 .”
Towards Unbounded Machine Unlearning (2023, background): “More recently, Guo et al. [2019], Izzo et al. [2021], Sekhari et al. [2021] built upon this framework and introduced unlearning methods that can provide theoretical guarantees under certain assumptions.”
Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning (2024, background): “Since its proposal by Cao & Yang (2015), machine unlearning has been extensively studied in the classification literature (Bourtoule et al., 2021; Golatkar et al., 2020; Ginart et al., 2019; Thudi et al., 2022; Izzo et al., 2021; Koh & Liang, 2017; Guo et al., 2019; Sekhari et al., 2021).”