Privacy-Preserving Machine Learning

The popularization of cloud computing and machine learning algorithms facilitates ranges of complex analytic services, such as medical or financial assessments. This allows a computationally-limited client to get predictions or classification results by paying for the analytic services. These services often involve sensitive data which should be kept private. Ideally, we hope for privacy-preserving machine learning services. The clients can learn the results of the model from a service provider without revealing their inputs and the results. Meanwhile, the trained model is kept confidential from the clients with as minimal leakage as possible. This keynote focuses on how cryptography can enable privacy-preserving machine learning services, in particular, decision tree evaluation.