Daniel A. Abolafia — accepted AI-conference papers

2019

Bayesian Deep Convolutional Networks with Many Channels are Gaussian Processes

Roman Novak, Lechao Xiao, Yasaman Bahri, Jaehoon Lee, Greg Yang, Jiri Hron +3

ICLR 2019poster

There is a previously identified equivalence between wide fully connected neural networks (FCNs) and Gaussian processes (GPs). This equivalence enables, for instance, test set predictions that would have resulted from a fully Bayesian, infinitely wide trained FCN to be computed without ever instanti…

Cited by 401 Source ↗PDF ↗Scholar ↗

2018

Sensitivity and Generalization in Neural Networks: an Empirical Study

Roman Novak, Yasaman Bahri, Daniel A. Abolafia, Jeffrey Pennington, Jascha Sohl-Dickstein

ICLR 2018poster

In practice it is often found that large over-parameterized neural networks generalize better than their smaller counterparts, an observation that appears to conflict with classical notions of function complexity, which typically favor smaller models. In this work, we investigate this tension betwee…

Cited by 545 Source ↗PDF ↗Scholar ↗