/Keras_IEBN

Unofficial Keras implementation of the paper Instance Enhancement Batch Normalization.

Primary LanguageJupyter NotebookMIT LicenseMIT

Instance Enhancement Batch Normalization: an Adaptive Regulator of Batch Noise

This repository is an unofficial Keras implementation of the paper Instance Enhancement Batch Normalization by Senwei Liang, Zhongzhan Huang, Mingfu Liang and Haizhao Yang.

The official Pytorch implementation is available here : https://github.com/gbup-group/IEBN

Introduction

Instance Enhancement Batch Normalization (IEBN) is an attention-based version of BN which recalibrates channel information of BN by a simple linear transformation, this can be used as a droppin replacement of standard BatchNormalization layer.

Usage

Please refer to the notebook for an usage example.