Acceleration of Convolutional Neural Network Using FFT-Based Split Convolutions

TL;DR

This paper introduces a novel FFT-based split convolution method for CNNs that reduces computational complexity and enhances efficiency, especially with small kernels, through input splitting and hardware implementation analysis.

Contribution

It proposes a new FFT domain processing technique for CNNs using input splitting to reduce complexity and improve efficiency, particularly for small kernels.

Findings

Reduced FFT computation complexity with input splitting

Improved efficiency through overlap-and-add redundancy reduction

Validated hardware implementation demonstrating performance gains

Abstract

Convolutional neural networks (CNNs) have a large number of variables and hence suffer from a complexity problem for their implementation. Different methods and techniques have developed to alleviate the problem of CNN's complexity, such as quantization, pruning, etc. Among the different simplification methods, computation in the Fourier domain is regarded as a new paradigm for the acceleration of CNNs. Recent studies on Fast Fourier Transform (FFT) based CNN aiming at simplifying the computations required for FFT. However, there is a lot of space for working on the reduction of the computational complexity of FFT. In this paper, a new method for CNN processing in the FFT domain is proposed, which is based on input splitting. There are problems in the computation of FFT using small kernels in situations such as CNN. Splitting can be considered as an effective solution for such issues aroused by small kernels. Using splitting redundancy, such as overlap-and-add, is reduced and, efficiency is increased. Hardware implementation of the proposed FFT method, as well as different analyses of the complexity, are performed to demonstrate the proper performance of the proposed method.