Machine Learning (ML) as-a-service (MLaaS) has brought much convenience to our daily lives. Recent services provided by the IT industries are growing rapidly including: Microsoft Azure Machine Learning Studio, AWS Machine Learning and Google Cloud Machine Learning Engine. However, these MLaaS are offered through cloud computing services which raises the potential of privacy leakage when personal data were used in the model development. How to preserve privacy as well as preventing abusive usage of sensitive personal data, while at the same time enjoy the convenience and knowledge brought by deep learning, becomes an important issue.
This talk aims to provide a broad overview over various security aspects in machine learning pipeline, including how security can be enhanced by applying machine learning to active authentication scheme, as well as security issues against attacks that use machine learning. Threat models such as model inversion attacks, membership inference attack, adversarial example attack, as well as remedies including differential privacy, cryptographic approaches, compressive privacy, as well as randomized smoothing, will be introduced.