Cryptology ePrint Archive: Report 2020/159

Privacy-preserving collaborative machine learning on genomic data using TensorFlow

Cheng Hong and Zhicong Huang and Wen-jie Lu and Hunter Qu and Li Ma and Morten Dahl and Jason Mancuso

Abstract: Machine learning (ML) methods have been widely used in genomic studies. However, genomic data are often held by different stakeholders (e.g. hospitals, universities, and healthcare companies) who consider the data as sensitive information, even though they desire to collaborate. To address this issue, recent works have proposed solutions using Secure Multi-party Computation (MPC), which train on the decentralized data in a way that the participants could learn nothing from each other beyond the final trained model. We design and implement several MPC-friendly ML primitives, including class weight adjustment and parallelizable approximation of activation function. In addition, we develop the solution as an extension to TF Encrypted (Dahl et al., 2018), enabling us to quickly experiment with enhancements of both machine learning techniques and cryptographic protocols while leveraging the advantages of TensorFlow’s optimizations. Our implementation compares favorably with state-ofthe-art methods, winning first place in Track IV of the iDASH2019 secure genome analysis competition. 1

Category / Keywords: implementation / Machine learning, MPC

Original Publication (in the same form): https://arxiv.org/abs/2002.04344

Date: received 12 Feb 2020

Contact author: vince hc at alibaba-inc com

Available format(s): PDF | BibTeX Citation

Note: Description of the winning solution at Track IV of iDASH competition 2019

Version: 20200213:132914 (All versions of this report)

Short URL: ia.cr/2020/159


[ Cryptology ePrint archive ]