Phillip Rieger (Technical University of Darmstadt), Alessandro Pegoraro (Technical University of Darmstadt), Kavita Kumari (Technical University of Darmstadt), Tigist Abera (Technical University of Darmstadt), Jonathan Knauer (Technical University of Darmstadt), Ahmad-Reza Sadeghi (Technical University of Darmstadt)

Split Learning (SL) is a distributed deep learning approach enabling multiple clients and a server to collaboratively train and infer on a shared deep neural network (DNN) without requiring clients to share their private local data. The DNN is partitioned in SL, with most layers residing on the server and a few initial layers and inputs on the client side. This configuration allows resource-constrained clients to participate in training and inference. However, the distributed architecture exposes SL to backdoor attacks, where malicious clients can manipulate local datasets to alter the DNN's behavior. Existing defenses from other distributed frameworks like Federated Learning are not applicable, and there is a lack of effective backdoor defenses specifically designed for SL.

We present SafeSplit, the first defense against client-side backdoor attacks in Split Learning (SL). SafeSplit enables the server to detect and filter out malicious client behavior by employing circular backward analysis after a client's training is completed, iteratively reverting to a trained checkpoint where the model under examination is found to be benign. It uses a two-fold analysis to identify client-induced changes and detect poisoned models. First, a static analysis in the frequency domain measures the differences in the layer's parameters at the server. Second, a dynamic analysis introduces a novel rotational distance metric that assesses the orientation shifts of the server's layer parameters during training. Our comprehensive evaluation across various data distributions, client counts, and attack scenarios demonstrates the high efficacy of this dual analysis in mitigating backdoor attacks while preserving model utility.

View More Papers

Starshields for iOS: Navigating the Security Cosmos in Satellite...

Jiska Classen (Hasso Plattner Institute, University of Potsdam), Alexander Heinrich (TU Darmstadt, Germany), Fabian Portner (TU Darmstadt, Germany), Felix Rohrbach (TU Darmstadt, Germany), Matthias Hollick (TU Darmstadt, Germany)

Read More

Cascading Spy Sheets: Exploiting the Complexity of Modern CSS...

Leon Trampert (CISPA Helmholtz Center for Information Security), Daniel Weber (CISPA Helmholtz Center for Information Security), Lukas Gerlach (CISPA Helmholtz Center for Information Security), Christian Rossow (CISPA Helmholtz Center for Information Security), Michael Schwarz (CISPA Helmholtz Center for Information Security)

Read More

Onion Franking: Abuse Reports for Mix-Based Private Messaging

Matthew Gregoire (University of North Carolina at Chapel Hill), Margaret Pierce (University of North Carolina at Chapel Hill), Saba Eskandarian (University of North Carolina at Chapel Hill)

Read More

Mysticeti: Reaching the Latency Limits with Uncertified DAGs

Kushal Babel (Cornell Tech & IC3), Andrey Chursin (Mysten Labs), George Danezis (Mysten Labs & University College London (UCL)), Anastasios Kichidis (Mysten Labs), Lefteris Kokoris-Kogias (Mysten Labs & IST Austria), Arun Koshy (Mysten Labs), Alberto Sonnino (Mysten Labs & University College London (UCL)), Mingwei Tian (Mysten Labs)

Read More