Gradient Normalization & Depth Based Decay For Deep Learning

TL;DR

This paper presents a simple yet effective gradient normalization and depth-based decay method that improves convergence times in deep neural networks across image classification and NLP tasks.

Contribution

It introduces a novel gradient normalization and decay technique that can be integrated with existing optimizers to enhance training efficiency.

Findings

Improved convergence time on DenseNet and ResNet for image classification

Enhanced training efficiency for LSTM in NLP tasks

Compatible with most current optimizers

Abstract

In this paper we introduce a novel method of gradient normalization and decay with respect to depth. Our method leverages the simple concept of normalizing all gradients in a deep neural network, and then decaying said gradients with respect to their depth in the network. Our proposed normalization and decay techniques can be used in conjunction with most current state of the art optimizers and are a very simple addition to any network. This method, although simple, showed improvements in convergence time on state of the art networks such as DenseNet and ResNet on image classification tasks, as well as on an LSTM for natural language processing tasks.