Performance Guaranteed Network Acceleration via High-Order Residual Quantization

TL;DR

This paper introduces a high-order residual binarization method for neural networks that improves approximation accuracy and maintains binary operation advantages, leading to better recognition performance.

Contribution

The paper proposes a novel high-order residual binarization scheme with recursive residual quantization and new binary filtering, enhancing accuracy without sacrificing efficiency.

Findings

Achieves higher recognition accuracy compared to previous binarization methods.

Maintains computational efficiency through binary operations.

Provides theoretical error bounds for the approximation.

Abstract

Input binarization has shown to be an effective way for network acceleration. However, previous binarization scheme could be regarded as simple pixel-wise thresholding operations (i.e., order-one approximation) and suffers a big accuracy loss. In this paper, we propose a highorder binarization scheme, which achieves more accurate approximation while still possesses the advantage of binary operation. In particular, the proposed scheme recursively performs residual quantization and yields a series of binary input images with decreasing magnitude scales. Accordingly, we propose high-order binary filtering and gradient propagation operations for both forward and backward computations. Theoretical analysis shows approximation error guarantee property of proposed method. Extensive experimental results demonstrate that the proposed scheme yields great recognition accuracy while being accelerated.