Abstract: We consider statistical inference tasks in a distributed setting where access to data samples is subjected to strict "local constraints," through a unified framework that captures communication limitations and (local) privacy constraints as special cases. We study estimation (learning) and goodness-of-fit (testing) for both discrete and high-dimensional distributions. Our goal is to understand how the sample complexity increases under the information constraints.
In this talk we will provide an overview of this field and a sample of some of our results. We will discuss the role of (public) randomness and interactivity in information-constrained inference, and make a case for thinking about randomness and interactivity as resources.
The work is part of a long-term ongoing collaboration with Clément Canonne (IBM Research) and Himanshu Tyagi (IISc), and includes works done with Cody Freitag (Cornell), Yanjun Han (Stanford), Yuhan Liu (Cornell), and Ziteng Sun (Cornell).
To watch the talk:
- Watching the live stream. At the announced start time of the talk (or a minute before), a live video stream will be available on our "next talk" page. Simply connect to the page and enjoy the talk. No webcam or registration is needed. Questions and comments during the talk are welcome (text only, unfortunately); simply post a comment below the live video stream on YouTube.
- Watching the recorded talk offline. The recorded talk will be made available shortly after the talk ends on our YouTube page. (Please leave a comment if you enjoyed it!)