Depth-wise Decomposition for Accelerating Separable Convolutions in Efficient Convolutional Neural Networks

TL;DR

This paper introduces a novel SVD-based depth-wise decomposition method to accelerate separable convolutions in CNNs, maintaining high accuracy while reducing inference latency, demonstrated on ShuffleNet V2 and ImageNet.

Contribution

It proposes a new decomposition technique using GSVD to expand regular convolutions into efficient depthwise separable convolutions with minimal accuracy loss.

Findings

Outperforms channel decomposition on all tested datasets.

Improves ShuffleNet V2 Top-1 accuracy by approximately 2%.

Effective on both synthetic and large-scale datasets.

Abstract

Very deep convolutional neural networks (CNNs) have been firmly established as the primary methods for many computer vision tasks. However, most state-of-the-art CNNs are large, which results in high inference latency. Recently, depth-wise separable convolution has been proposed for image recognition tasks on computationally limited platforms such as robotics and self-driving cars. Though it is much faster than its counterpart, regular convolution, accuracy is sacrificed. In this paper, we propose a novel decomposition approach based on SVD, namely depth-wise decomposition, for expanding regular convolutions into depthwise separable convolutions while maintaining high accuracy. We show our approach can be further generalized to the multi-channel and multi-layer cases, based on Generalized Singular Value Decomposition (GSVD) [59]. We conduct thorough experiments with the latest ShuffleNet V2 model [47] on both random synthesized dataset and a large-scale image recognition dataset: ImageNet [10]. Our approach outperforms channel decomposition [73] on all datasets. More importantly, our approach improves the Top-1 accuracy of ShuffleNet V2 by ~2%.