William (Bill) Peebles

email: peebles [at] berkeley [dot] edu

RESEARCH UPDATES

Check out our recent work on unsupervised disentanglement in deep generative models.

THE HESSIAN PENALTY

About Me

I'm a second-year PhD student at UC Berkeley advised by Alyosha Efros. Previously, I did my undergrad at MIT where I worked with Antonio Torralba. My research currently focuses on unsupervised learning, with an emphasis on deep generative models for images and videos. I'm supported by the National Science Foundation's Graduate Research Fellowship Program.

News

[Aug 2020] Code for the Hessian Penalty has been released in PyTorch and TensorFlow here.
[May 2020] I'm interning at Adobe Research for the summer.

Publications

The Hessian Penalty: A Weak Prior for Unsupervised Disentanglement
William Peebles, John Peebles, Jun-Yan Zhu, Alexei Efros, Antonio Torralba
ECCV 2020 (Spotlight)
video · overview in 90 seconds · project page · paper · github

GAN reconstructions of images versus the original images. Seeing What a GAN Cannot Generate
David Bau, Jun-Yan Zhu, Jonas Wulff, William Peebles, Hendrik Strobelt, Bolei Zhou, Antonio Torralba
ICCV 2019 (Oral)
demo · paper · overview · github

A natural image of a building is modified by adding trees and domes. Semantic Photo Manipulation with a Generative Image Prior
David Bau, Hendrik Strobelt, William Peebles, Jonas Wulff, Bolei Zhou, Jun-Yan Zhu, Antonio Torralba
SIGGRAPH 2019
demo · paper · overview