[SEMINAR REGISTRATION] Benign Overfitting - VinAI Research
Seminar Series: Benign Overfitting
Speaker: Peter Bartlett - Associate Director, Simons Institute - Professor, EECS and Statistics, University of California, Berkeley
Date: 10.00 am - 11.30 am, Tuesday, November 12, 2019
Location: COGO Co-Working Space, 4th Floor, Sun Grand City Ancora Residence, No.3 Luong Yen Street, Hai Ba Trung District, Hanoi

About the Speaker:

Peter Bartlett is the Associate Director of the Simons Institute for the Theory of Computing and professor in EECS and Statistics at the University of California at Berkeley. His research interests include machine learning and statistical learning theory, and he is the co-author, with Martin Anthony, of the book Neural Network Learning: Theoretical Foundations. He has been Institute of Mathematical Statistics Medallion Lecturer, winner of the Malcolm McIntosh Prize for Physical Scientist of the Year, Australian Laureate Fellow, Fellow of the IMS, Fellow of the ACM, and Fellow of the Australian Academy of Science.


Classical theory that guides the design of nonparametric prediction methods like deep neural networks involves a tradeoff between the fit to the training data and the complexity of the prediction rule. Deep learning seems to operate outside the regime where these results are informative, since deep networks can perform well even with a perfect fit to noisy training data. We investigate this phenomenon of 'benign overfitting' in the simplest setting, that of linear prediction. We give a characterization of linear regression problems for which the minimum norm interpolating prediction rule has near-optimal prediction accuracy. The characterization is in terms of two notions of effective rank of the data covariance. It shows that massive overparameterization is essential: the number of directions in parameter space that are unimportant for prediction must significantly exceed the sample size. We discuss implications for deep networks and for robustness to adversarial examples.

Joint work with Phil Long, Gábor Lugosi, and Alex Tsigler.
Your full name *
Your answer
Your email *
Your answer
Your phone number *
Your answer
Institution/Company *
Your answer
Your current position *
Your answer
Do you have any question for our Speaker?
Your answer
Never submit passwords through Google Forms.
This content is neither created nor endorsed by Google. Report Abuse - Terms of Service - Privacy Policy