SIGuard: Guarding Secure Inference with Post Data Privacy
Xinqian Wang
Network and Distributed System Security (NDSS) Symposium 2025 · Day 3 · Membership Inference
The proliferation of machine learning as a service (MLaaS) has revolutionized how intelligence is consumed, offering sophisticated prediction capabilities through cloud-hosted neural networks. While this paradigm provides immense value, particularly in sensitive domains like medical imaging, it introduces significant privacy challenges. Traditional MLaaS workflows involve users submitting data to a cloud provider for inference, receiving a prediction and often a confidence vector in return. This interaction, however, exposes both the user's sensitive data and the model owner's valuable intellectual property to potential privacy breaches.
AI review
Legitimate academic research on a real gap — output privacy in MPC-based secure inference — with a concrete novel contribution in the fixed-iteration loop design to prevent side-channel leakage in colluding threat models. Solid work for a PhD student at NDSS, but it's a niche extension of an existing defense (Mamgard) rather than a fundamental advance, and the 50x efficiency claim needs more scrutiny than a conference talk summary can provide.