Get our free extension to see links to code for papers anywhere online!

Add to Chrome

Add to Firefox

Picture for Boris Hanin

Boris Hanin

Bayesian Interpolation with Deep Linear Networks

Jan 02, 2023
Boris Hanin, Alexander Zlokapa

Figure 1 for Bayesian Interpolation with Deep Linear Networks

Figure 2 for Bayesian Interpolation with Deep Linear Networks

Figure 3 for Bayesian Interpolation with Deep Linear Networks

Access Paper or Ask Questions

Correlation Functions in Random Fully Connected Neural Networks at Finite Width

Apr 03, 2022
Boris Hanin

* 88p

Access Paper or Ask Questions

Ridgeless Interpolation with Shallow ReLU Networks in $1D$ is Nearest Neighbor Curvature Extrapolation and Provably Generalizes on Lipschitz Functions

Sep 27, 2021
Boris Hanin

Figure 1 for Ridgeless Interpolation with Shallow ReLU Networks in $1D$ is Nearest Neighbor Curvature Extrapolation and Provably Generalizes on Lipschitz Functions

Figure 2 for Ridgeless Interpolation with Shallow ReLU Networks in $1D$ is Nearest Neighbor Curvature Extrapolation and Provably Generalizes on Lipschitz Functions

Figure 3 for Ridgeless Interpolation with Shallow ReLU Networks in $1D$ is Nearest Neighbor Curvature Extrapolation and Provably Generalizes on Lipschitz Functions

Figure 4 for Ridgeless Interpolation with Shallow ReLU Networks in $1D$ is Nearest Neighbor Curvature Extrapolation and Provably Generalizes on Lipschitz Functions

Access Paper or Ask Questions

Random Neural Networks in the Infinite Width Limit as Gaussian Processes

Jul 04, 2021
Boris Hanin

* 26p

Access Paper or Ask Questions

The Principles of Deep Learning Theory

Jun 18, 2021
Daniel A. Roberts, Sho Yaida, Boris Hanin

* 451 pages, to be published by Cambridge University Press

Access Paper or Ask Questions

Deep ReLU Networks Preserve Expected Length

Feb 21, 2021
Boris Hanin, Ryan Jeong, David Rolnick

Figure 1 for Deep ReLU Networks Preserve Expected Length

Figure 2 for Deep ReLU Networks Preserve Expected Length

Figure 3 for Deep ReLU Networks Preserve Expected Length

Figure 4 for Deep ReLU Networks Preserve Expected Length

* 25 pages, 5 figures

Access Paper or Ask Questions

Data augmentation as stochastic optimization

Oct 21, 2020
Boris Hanin, Yi Sun

Figure 1 for Data augmentation as stochastic optimization

Figure 2 for Data augmentation as stochastic optimization

* 25 pages

Access Paper or Ask Questions

Finite Depth and Width Corrections to the Neural Tangent Kernel

Sep 13, 2019
Boris Hanin, Mihai Nica

Figure 1 for Finite Depth and Width Corrections to the Neural Tangent Kernel

Figure 2 for Finite Depth and Width Corrections to the Neural Tangent Kernel

* 27 pages, 2 figures, comments welcome

Access Paper or Ask Questions