Joint Regularization on Activations and Weights for Efficient Neural Network Pruning

Abstract

With the rapid scaling up of deep neural networks (DNNs), extensive research studies on network model compression such as weight pruning have been performed for improving deployment efficiency. This work aims to advance the compression beyond the weights to neuron activations. We propose the joint regularization technique which simultaneously regulates the distribution of weights and activations. By distinguishing and leveraging the significance difference among neuron responses and connections during learning, the jointly pruned network, namely \textit{JPnet}, optimizes the sparsity of activations and weights for improving execution efficiency. The derived deep sparsification of JPnet reveals more optimization space for the existing DNN accelerators dedicated for sparse matrix operations. We thoroughly evaluate the effectiveness of joint regularization through various network models with different activation functions and on different datasets. With $0.4\%$ degradation constraint on inference accuracy, a JPnet can save $72.3\%\sim 98.8\%$ of computation cost compared to the original dense models, with up to $5.2\times$ and $12.3\times$ reductions in activation and weight numbers, respectively.

References in corpus (11)

• ADADELTA: An Adaptive Learning Rate Method
• Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
• Wide Residual Networks
• Pruning Convolutional Neural Networks for Resource Efficient Inference
• A Systematic DNN Weight Pruning Framework using Alternating Direction Method of Multipliers
• Variational Dropout Sparsifies Deep Neural Networks
• Winner-Take-All Autoencoders
• Rethinking the Smaller-Norm-Less-Informative Assumption in Channel Pruning of Convolution Layers
• Faster CNNs with Direct Sparse Convolutions and Guided Pruning
• Structured Bayesian Pruning via Log-Normal Multiplicative Noise
• Compressing Neural Networks using the Variational Information Bottleneck