Hidde Lycklama
March 15, 2024 at 11:00 AM on Zoom / Soda Hall
Holding Secrets Accountable: Auditing Privacy-Preserving Machine Learning
Abstract:
Recent advancements in privacy-preserving machine learning are paving the way to extend the benefits of ML to highly sensitive data that, until now, have been hard to utilize due to privacy concerns and regulatory constraints. Simultaneously, there is a growing emphasis on enhancing the transparency and accountability of machine learning, including the ability to audit ML deployments. Despite both ML transparency and PPML being examined predominately in isolation, the need to study their combination is increasingly being recognized by works such as Holmes and Cerebro.
In this work, we introduce Arc, an MPC framework for auditing privacy-preserving machine learning. Our system is highly modular and supports a wide range of efficient PPML approaches and auditing functions. At the core of our framework is a new protocol for efficiently verifying MPC inputs against succinct commitments at scale. We evaluate the performance of our framework when instantiated with our consistency protocol and compare it to an approach based on hashes and the homomorphic-commitment-based approach from Cerebro, demonstrating that it is up to 10^4 times faster and up to 10^6 times more concise.
Bio: Hidde Lycklama is a 3rd year PhD student at ETH Zurich at the Privacy-Preserving Systems Lab supervised by Anwar Hithnawi. His research focuses on the robustness and accountability of secure learning systems. Designing systems that balance privacy and transparency through the application of cryptography is one of his key interests.