Shun John IwaseinBetween Real and IdealBatch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift 4/1004th thesis is about Batch Normalization (a.k.a BN) which allows us to use high learning rates and be less careful about initialization…Apr 13, 2018Apr 13, 2018
Shun John IwaseinBetween Real and IdealExtremely Large Minibatch SGD: Training ResNet-50 on ImageNet in 15 Minutes 3/1003rd thesis is about Rest-50 training with using extremely large minibatch SGD.Apr 10, 2018Apr 10, 2018
Shun John IwaseinBetween Real and IdealDistributed Second-Order Optimization using Kronecker-Factored Approximations 2/1002nd thesis is about speeding up training with using K-FAC method instead of SGD.Apr 10, 2018Apr 10, 2018
Shun John IwaseinBetween Real and IdealImagenet Training in Minutes 1/100Just I started challenge to read 100 theses until May 6th. It’s just note to myself. Feel free to quote and make a link to this article.Apr 10, 2018Apr 10, 2018
Shun John IwaseinBetween Real and IdealAbout my dreamFrom this April, I got in the master course of computer science and became a member of the Yokota Lab in Tokyo Institute of Technology…Apr 8, 2018Apr 8, 2018