Speeding-up Convolutional Neural Networks Using Fine-tuned CP-Decomposition

TL;DR

This paper introduces a two-step method using tensor CP-decomposition and fine-tuning to accelerate convolutional neural networks, achieving significant speedups with minimal accuracy loss.

Contribution

The authors present a novel approach combining tensor decomposition with discriminative fine-tuning to efficiently speed up CNN layers.

Findings

Achieved 8.5x CPU speedup on a character classification CNN with 1% accuracy drop.

Speeded up AlexNet's second convolution layer by 4x with 1% increase in top-5 error.

Method is competitive with previous approaches, balancing speed and accuracy.

Abstract

We propose a simple two-step approach for speeding up convolution layers within large convolutional neural networks based on tensor decomposition and discriminative fine-tuning. Given a layer, we use non-linear least squares to compute a low-rank CP-decomposition of the 4D convolution kernel tensor into a sum of a small number of rank-one tensors. At the second step, this decomposition is used to replace the original convolutional layer with a sequence of four convolutional layers with small kernels. After such replacement, the entire network is fine-tuned on the training data using standard backpropagation process. We evaluate this approach on two CNNs and show that it is competitive with previous approaches, leading to higher obtained CPU speedups at the cost of lower accuracy drops for the smaller of the two networks. Thus, for the 36-class character classification CNN, our approach obtains a 8.5x CPU speedup of the whole network with only minor accuracy drop (1% from 91% to 90%). For the standard ImageNet architecture (AlexNet), the approach speeds up the second convolution layer by a factor of 4x at the cost of $1\%$ increase of the overall top-5 classification error.