Guest Talk - Abhishek Shetty

— 11:00am

Location:
In Person - Gates Hillman 8102

Speaker:
ABHISHEK SHETTY , FODSI Postdoctoral Fellow, Massachussets Institute of Technology , Catherine M. and James E. Allchin Early-Career Assistant Professor in Computer Science, Georgia Institute of Technology (incoming)
https://ashettyv.github.io/

When does the past predict the future: Recent perspectives from Smoothed Online Learning, Abstention and Optimal PAC Learning

How one makes use of data that has been collected in the past to inform us about decisions in the future is perhaps one of the most fundamental questions in computer science. Given its fundamental nature, several frameworks have been devised to understand this question, but a key assumption lies at the heart of these is independence. Through the years several frameworks have been proposed to circumvent this requirement, many of these suffer from intractability either statistically or computationally. 

In this talk, we will survey a few recent techniques that have been proposed to handle these issues. Most of the talk will focus on smoothed online learning which provides tools to design statistically and computationally efficient algorithms even when the data distribution changes over time. Further, we will see how this ties together with the notion of abstention in online learning, which too leads to surprising statistically efficient algorithms. Time permitting, we will briefly touch upon how analyzing generalization under dependent data sheds light on optimal algorithms even when the data is independent.   

— 

Abhishek Shetty is an incoming Catherine M. and James E. Allchin Early-Career Assistant Professor in the School of Computer Science at Georgia Tech and is currently FODSI Postdoctoral Fellow at MIT, hosted by Sasha Rakhlin, Ankur Moitra and Costis Daskalakis. He graduated from the department of EECS at UC Berkeley advised by Nika Haghtalab. His interests lie at the intersection of machine learning, theoretical computer science and statistics, especially aimed at understanding how fundamental algorithmic techniques can contribute to modern machine learning. His research has been awarded with the Apple AI/ML fellowship and the American Statistical association SCGS best student paper. Faculty Host:  Andrej Risteski

Add event to Google
Add event to iCal