Secure Quantized Training for Deep Learning
Published in ICML, 2022
This paper describes our implementation of deep learning training in MP-SPDZ.
Download here
Publications
This is a selection of recent publications. You can find a complete list on Google Scholar and Semantic Scholar
Fantastic Four: Honest-Majority Four-Party Secure Computation With Malicious Security
Published in USENIX Security, 2021
This paper describes an efficient four-party protocol with active security and applications to machine learning.
Download here
MP-SPDZ: A versatile framework for multi-party computation
Published in ACM CCS, 2020
This paper describes the protocols and the design of MP-SPDZ.
Download here
Improved Primitives for MPC over Mixed Arithmetic-Binary Circuits
Published in IACR CRYPTO, 2020
This paper proposes edaBits, a general way of switching between arithmetic and binary computation in MPC.
Download here
Secure Evaluation of Quantized Neural Networks
Published in PETS, 2020
This paper describes an implementation of MobileNets in secure computation.
Download here
New Primitives for Actively-Secure MPC over Rings with Applications to Private Machine Learning
Published in IEEE Security & Privacy, 2019
This paper describes an implementation of SPDZ2k, a protocol with malicious security based on oblivious transfer.
Download here