ESCAPED: Efficient Secure and Private Dot Product Framework for Kernel-based Machine Learning Algorithms with Applications in Healthcare

Ali Burak Ünal,Mete Akgün,Nico Pfeifer

DOI: https://doi.org/10.48550/arXiv.2012.02688

2020-12-04

Abstract:To train sophisticated machine learning models one usually needs many training samples. Especially in healthcare settings these samples can be very expensive, meaning that one institution alone usually does not have enough on its own. Merging privacy-sensitive data from different sources is usually restricted by data security and data protection measures. This can lead to approaches that reduce data quality by putting noise onto the variables (e.g., in $\epsilon$-differential privacy) or omitting certain values (e.g., for $k$-anonymity). Other measures based on cryptographic methods can lead to very time-consuming computations, which is especially problematic for larger multi-omics data. We address this problem by introducing ESCAPED, which stands for Efficient SeCure And PrivatE Dot product framework, enabling the computation of the dot product of vectors from multiple sources on a third-party, which later trains kernel-based machine learning algorithms, while neither sacrificing privacy nor adding noise. We evaluated our framework on drug resistance prediction for HIV-infected people and multi-omics dimensionality reduction and clustering problems in precision medicine. In terms of execution time, our framework significantly outperforms the best-fitting existing approaches without sacrificing the performance of the algorithm. Even though we only show the benefit for kernel-based algorithms, our framework can open up new research opportunities for further machine learning models that require the dot product of vectors from multiple sources.

Machine Learning,Cryptography and Security

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to efficiently calculate the vector dot product from multiple data sources on the premise of protecting privacy, in order to support the application of kernel - based machine learning algorithms in the medical and health field. Specifically, due to the high sensitivity of medical data, there are many legal and ethical obstacles to directly sharing raw data between different institutions. This leads to a decline in data quality (for example, by adding noise or omitting certain values to protect privacy) or the calculation process becomes very time - consuming (for example, using encryption methods). To solve these problems, the authors propose the ESCAPED framework, that is, an efficient, secure, and private dot - product framework, which aims to enable third - parties to calculate the dot product of multi - source vectors without revealing the data privacy of each participating party, and use it to train kernel - based machine learning models.

The ESCAPED framework is especially suitable for scenarios where information needs to be obtained from multiple data sources, such as predicting drug resistance in HIV - infected people, and performing dimensionality reduction and clustering analysis of multi - omics data in precision medicine. Experiments have proven that the ESCAPED framework is significantly superior to the existing best solutions in terms of execution time, while maintaining algorithm performance and without sacrificing data privacy. In addition, although the paper mainly shows the application of ESCAPED in kernel - based algorithms, this framework can also open up new research opportunities for other machine learning models that need to calculate the multi - source vector dot product.