Abstract:

Machine learning-based applications provide a lot of convenience but rely on our personal data to work well. Can we get the benefits of using them without disclosing sensitive information about ourselves? We present cryptographic protocols based on Secure Multiparty Computation (SMC) that let machine learning-backed applications provide their services without demanding our privacy in return. The protocols run on Microsoft Azure virtual machines in a machine-learning-as-a-service setup while keeping the data as well as the model parameters encrypted. We present use cases of privacy-preserving classification of personal text messages and privacy-preserving recognition of people’s emotions in videos, both with state-of-the-art deep learning models.

Case Study Summary:

  • The scientific problem we tackled:
    Machine learning-based applications provide a lot of convenience but rely on our personal data to work well. Can we get the benefits of using them without disclosing sensitive information about ourselves?
  • The computational methods we used:
    We use cryptographic protocols based on Secure Multiparty Computation (SMC) that let machine learning-backed applications provide their services without demanding our privacy in return. Use cases include privacy-preserving classification of personal text messages and privacy-preserving recognition of people’s emotions in videos, both with state-of-the-art deep learning models.
  • The cloud resources we used:
    The protocols run on Microsoft Azure virtual machines in a machine-learning-as-a-service setup while keeping the data as well as the model parameters encrypted.
  • The differences we’ve observed between locally-provided and cloud-provided resources:
    Benchmarking our protocols in the cloud is important in the research group and in the broader research community. It enables reproducibility of research results, and consistent apples-to-apples comparisons of the efficiency and scalability of different cryptographic protocols on the same kind of hardware.

Author Bio:

Martine De Cock is a Professor at the School of Engineering and Technology, University of Washington–Tacoma, USA and a guest professor at Ghent University, Belgium. She has over 190 peer-reviewed publications in international journals and conferences on artificial intelligence, machine learning, information retrieval, web intelligence, and logic programming. Her current research interests are privacy-preserving machine learning (PPML) and machine learning for cybersecurity. She holds a patent on cryptographically secure machine learning, and her team won Track IV of iDASH2019, the most significant competition in privacy-preserving analysis of genomic data in the world.

For further information:

RRoCCET21 is a conference that was held virtually by CloudBank from August 10th through 12th, 2021. Its intention is to inspire you to consider utilizing the cloud in your research, by way of sharing the success stories of others. We hope the proceedings, of which this case study is a part, give you an idea of what is possible and act as a “recipe book” for mapping powerful computational resources onto your own field of inquiry.

Tags