Abstract
Nowadays, as data becomes increasingly complex and distributed, data analysesoften involve several related datasets that are stored on different servers andprobably owned by different stakeholders. While there is an emerging need toprovide these stakeholders with a full picture of their data under a globalcontext, conventional visual analytical methods, such as dimensionalityreduction, could expose data privacy when multi-party datasets are fused into asingle site to build point-level relationships. In this paper, we reformulatethe conventional t-SNE method from the single-site mode into a securedistributed infrastructure. We present a secure multi-party scheme for jointt-SNE computation, which can minimize the risk of data leakage. Aggregatedvisualization can be optionally employed to hide disclosure of point-levelrelationships. We build a prototype system based on our method, SMAP, tosupport the organization, computation, and exploration of secure jointembedding. We demonstrate the effectiveness of our approach with three casestudies, one of which is based on the deployment of our system in real-worldapplications.