Sampled Softmax with Random Fourier Features
Ankit Singh Rawat, Jiecao Chen, Felix Yu, Ananda Theertha Suresh, and Sanjiv Kumar
Google Research, New York
{ankitsrawat, chenjiecao, felixyu, theertha, sanjivk}@google.com
Abstract
The computational cost of training with softmax cross entropy loss grows linearly
with the number of classes. For the settings where a large number of classes
are involved, a common method to speed u ...


雷达卡


京公网安备 11010802022788号







