Manticore: Efficient Framework for Scalable Secure Multiparty Computation Protocols

We propose a novel MPC framework, Manticore, in the multiparty setting, with full threshold and semi-honest security model, supporting a combination of real number arithmetic (arithmetic shares), Boolean arithmetic (Boolean shares) and garbled circuits (Yao shares). In contrast to prior work [34,32], Manticore never overflows, an important feature for machine learning applications. It achieves this without compromising efficiency or security. Compared to other overflow-free recent techniques such as MP-SPDZ [17] that convert arithmetic to Boolean shares, we introduce a novel highly efficient modular lifting/truncation method that stays in the arithmetic domain. We revisit some of the basic MPC operations such as real-valued polynomial evaluation, division, logarithms, exponentials and comparisons by employing our modular lift in combination with existing efficient conversions between arithmetic, Boolean and Yao shares. Furthermore, we provide a highly efficient and scalable implementation supporting logistic regression models with realworld training data sizes and high numerical precision through PCA and blockwise variants (for memory and runtime optimizations). On a dataset of 50 million rows and 50 columns distributed among two players, it completes in one day with at least 10 decimal digits of precision. Our logistic regression solution placed first at Track 3 of the annual iDASH’2020 Competition. Finally, we mention a novel oblivious sorting algorithm built using Manticore.