Peter Jin

cuw@rrpf.orexryrl.rqh

I’m a Computer Science PhD student in the ASPIRE, BAIR, BDD labs, where I’m part of Kurt Keutzer’s group. My current work is on stochastic optimization methods for deep learning, including parallel/distributed learning and new algorithms. I’m also interested in all things related to Monte Carlo tree search and GPUs.

I received my AB in Physics from Princeton University in 2012.

Preprints

Spatially Parallel Convolutions

Peter Jin, Boris Ginsburg, and Kurt Keutzer
ICLR 2018 Workshop Track
[openreview]

Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions

Bichen Wu, Alvin Wan, Xiangyu Yue, Peter Jin, Sicheng Zhao, Noah Golmant, Amir Gholaminejad, Joseph Gonzalez, and Kurt Keutzer
CVPR 2018
[arxiv]

Regret Minimization for Partially Observable Deep Reinforcement Learning

Peter Jin, Sergey Levine, and Kurt Keutzer
NIPS 2017 Deep RL Symposium, ICLR 2018 Workshop Track
[arxiv]

SqueezeDet: Unified, Small, Low Power Fully Convolutional Neural Networks for Real-Time Object Detection for Autonomous Driving

Bichen Wu, Forrest Iandola, Peter Jin, and Kurt Keutzer
CVPR Embedded Vision Workshop 2017
[arxiv]

How to scale distributed deep learning? [a.k.a. the gossiping SGD paper]

Peter Jin, Qiaochu Yuan, Forrest Iandola, and Kurt Keutzer
NIPS ML Systems Workshop 2016
[arxiv]

Convolutional Monte Carlo Rollouts in Go

Peter Jin and Kurt Keutzer
CG 2016 Neural Networks Workshop
[arxiv]