Personal tools

Learning Towards Better Accuracy and Privacy

Liyao Xiang, University of Toronto

  • Computer Science Seminar
  • Colloquium
When Fri, Jan 26, 2018
from 03:10 PM to 04:00 PM
Where Ritter Hall 115
Add event to calendar vCal
This talk starts with a real-world issue: to provide indoor localization services to satisfy contextual and ephemeral needs, e.g., at conferences or exhibitions events. As such, the costs and requirements of providing the services need to be minimal. We design, implement, and evaluate Tack, a new mobile application framework that uses a combination of known landmark locations, contacts over Bluetooth Low Energy, crowdsourcing, and dead-reckoning to estimate and refine user locations. At its core, an inference algorithm is designed to run on mobile devices to make the estimation more accurate.
Accuracy and privacy pose as a pair of contradictory requirements in machine learning frameworks -- stricter privacy guarantee is always achieved with degraded learning accuracy -- and such degradation is even worse with deep learning. We found the fundamental cause is that a loose characterization of utility and privacy leads to over-distortion of the model. By recognizing the accuracy-privacy tradeoff as a utility maximization problem subject to a set of privacy constraints, we lower-bounds the distortion, and significantly improves the learning accuracy as compared to the state-of-the-art under the same privacy guarantee. 
« May 2018 »