j201 / keras-ste-layers

STE (stochastically trained ensemble) layer implementation for Keras

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

STE Layers for Keras

Implementation of STE (stochastically trained ensemble) layers as introduced in arXiv:1911.09669.

Usage

STE layer code and full documentation is in ste.py. The only export is the STE class, which is more or less a drop-in replacement for the Keras Dense layer class, but with additional arguments for STE layers.

See example.py for a full example of using STE layers in LeNet-5.

About

STE (stochastically trained ensemble) layer implementation for Keras


Languages

Language:Python 100.0%