Passive Inference Attacks on Split Learning via Adversarial Regularization
This work addresses security vulnerabilities in Split Learning for privacy-sensitive applications, representing an incremental improvement over existing attacks.
The paper tackles the problem of passive inference attacks on Split Learning by introducing SDAR, a framework that uses adversarial regularization to reconstruct private client data. The result is significantly superior attack performance, achieving less than 0.025 mean squared error for feature reconstruction and over 98% label inference accuracy on CIFAR-10.
Split Learning (SL) has emerged as a practical and efficient alternative to traditional federated learning. While previous attempts to attack SL have often relied on overly strong assumptions or targeted easily exploitable models, we seek to develop more capable attacks. We introduce SDAR, a novel attack framework against SL with an honest-but-curious server. SDAR leverages auxiliary data and adversarial regularization to learn a decodable simulator of the client's private model, which can effectively infer the client's private features under the vanilla SL, and both features and labels under the U-shaped SL. We perform extensive experiments in both configurations to validate the effectiveness of our proposed attacks. Notably, in challenging scenarios where existing passive attacks struggle to reconstruct the client's private data effectively, SDAR consistently achieves significantly superior attack performance, even comparable to active attacks. On CIFAR-10, at the deep split level of 7, SDAR achieves private feature reconstruction with less than 0.025 mean squared error in both the vanilla and the U-shaped SL, and attains a label inference accuracy of over 98% in the U-shaped setting, while existing attacks fail to produce non-trivial results.