Machine learning unlearning speeds up security data removal in models
Correlation-Guided Fast Machine Unlearning via Hessian Analysis
Machine LearningArtificial Intelligence
Summary
Machine learning models sometimes need to forget specific training data, especially if that data is wrong or sensitive. This paper introduces a fast way to make models forget such data by finding related data points and updating the model's parameters using a new formula. This method is much quicker than existing ones and keeps the model accurate. It works well for security systems like spam filters and intrusion detection.
What this means in practice
- •For network security teams: Remove compromised or adversarial data quickly from machine learning models to maintain security systems without major downtime.
- •For privacy compliance professionals: Implement machine learning model updates that effectively erase user data to meet data deletion regulations like GDPR's right to be forgotten.
Authors
Ayushi Thakur, Ruchir Gupta, Amit Kumar Jaiswal, Prayag Tiwari
Abstract
The increasing adoption of machine learning in network and distributed security systems has created an urgent need for mechanisms that can selectively and efficiently remove the influence of specific training data to eliminate compromised or adversarial data points from production models. Privacy regulations such as GDPR's \emph{right to be forgotten} also pose similar requirements. However, existing approximate unlearning techniques remain computationally prohibitive for deployment in real-world security systems, as they require repeated expensive Hessian-inverse-vector computations for each data point removal, creating a bottleneck when processing multiple related requests in scenarios such as intrusion detection systems, spam filters, and threat intelligence platforms. Thus, we introduce a computationally efficient unlearning framework that identifies correlated data points in the training set and applies a theoretically derived closed-form parameter update rule, achieving an $82\times$ wall-clock speedup over standard influence function unlearning while preserving model utility with a $10^{-2}$ improvement in accuracy over state-of-the-art baselines. Our method establishes theoretical guarantees and ensures numerical stability through Hessian damping. Our evaluation across seven diverse dataset architecture combinations, including large-scale CIFAR-100 with ResNet-50, demonstrates superior forgetting effectiveness, with membership inference attack success rates of 0.660 and tug-of-war scores of 0.950.