Improving performance of recurrent neural network with relu nonlinearity

Abstract

In recent years significant progress has been made in successfully training recurrent neural networks (RNNs) on sequence learning problems involving long range temporal dependencies. The progress has been made on three fronts: (a) Algorithmic improvements involving sophisticated optimization techniques, (b) network design involving complex hidden layer nodes and specialized recurrent layer connections and (c) weight initialization methods. In this paper, we focus on recently proposed weight initialization with identity matrix for the recurrent weights in a RNN. This initialization is specifically proposed for hidden nodes with Rectified Linear Unit (ReLU) non linearity. We offer a simple dynamical systems perspective on weight initialization process, which allows us to propose a modified weight initialization strategy. We show that this initialization technique leads to successfully training RNNs composed of ReLUs. We demonstrate that our proposal produces comparable or better solution for three toy problems involving long range temporal structure: the addition problem, the multiplication problem and the MNIST classification problem using sequence of pixels. In addition, we present results for a benchmark action recognition problem.

10 pages 6 figures; under consideration for publication with ICLR 2016

References in corpus (8)

• LSTM: A Search Space Odyssey
• UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild
• On the difficulty of training Recurrent Neural Networks
• Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
• Unifying Visual-Semantic Embeddings with Multimodal Neural Language Models
• A Simple Way to Initialize Recurrent Networks of Rectified Linear Units
• Beyond Short Snippets: Deep Networks for Video Classification
• Random Walk Initialization for Training Very Deep Feedforward Networks

Cited by in corpus (11)

• Independently Recurrent Neural Network (IndRNN): Building A Longer and Deeper RNN
• Identifying nonlinear dynamical systems via generative recurrent neural networks with applications to fMRI
• Deep Independently Recurrent Neural Network (IndRNN)
• On the Initialization of Long Short-Term Memory Networks
• Identifying nonlinear dynamical systems with multiple time scales and long-range dependencies
• On the Curse of Memory in Recurrent Neural Networks: Approximation and Optimization Analysis
• Noisy Recurrent Neural Networks
• MomentumRNN: Integrating Momentum into Recurrent Neural Networks
• A Spatio-temporal Prediction Methodology Based on Deep Learning and Real Wi-Fi Measurements
• RNNs Evolving on an Equilibrium Manifold: A Panacea for Vanishing and Exploding Gradients?
• RNN Training along Locally Optimal Trajectories via Frank-Wolfe Algorithm