P
Preetum Nakkiran
Researcher at Harvard University
Publications - 55
Citations - 2212
Preetum Nakkiran is an academic researcher from Harvard University. The author has contributed to research in topics: Computer science & Artificial neural network. The author has an hindex of 15, co-authored 40 publications receiving 1128 citations. Previous affiliations of Preetum Nakkiran include Google & University of California, Berkeley.
Papers
More filters
Proceedings Article
Deep Double Descent: Where Bigger Models and More Data Hurt
TL;DR: The notion of model complexity allows us to identify certain regimes where increasing the number of train samples actually hurts test performance, and defines a new complexity measure called the effective model complexity and conjecture a generalized double descent with respect to this measure.
Journal ArticleDOI
A Discussion of 'Adversarial Examples Are Not Bugs, They Are Features'
Logan Engstrom,Justin Gilmer,Gabriel Goh,Dan Hendrycks,Andrew Ilyas,Aleksander Madry,Reiichiro Nakano,Preetum Nakkiran,Shibani Santurkar,Brandon Tran,Dimitris Tsipras,Eric Wallace +11 more
Proceedings ArticleDOI
Having your cake and eating it too: jointly optimal erasure codes for I/O, storage and network-bandwidth
TL;DR: This paper designs erasure codes that are simultaneously optimal in terms of I/O, storage, and network bandwidth, and builds on top of a class of powerful practical codes, called the product-matrix-MSR codes.
Posted Content
SGD on Neural Networks Learns Functions of Increasing Complexity.
Preetum Nakkiran,Gal Kaplun,Dimitris Kalimeris,Tristan Yang,Benjamin L. Edelman,Fred Zhang,Boaz Barak +6 more
TL;DR: Key to the work is a new measure of how well one classifier explains the performance of another, based on conditional mutual information, which can be helpful in explaining why SGD-learned classifiers tend to generalize well even in the over-parameterized regime.
Posted Content
Deep Double Descent: Where Bigger Models and More Data Hurt
TL;DR: In this paper, the authors show that a variety of modern deep learning tasks exhibit a "double-descent" phenomenon where, as we increase model size, performance first gets worse and then gets better.