Performance Guaranteed Network Acceleration via High-Order Residual Quantization
Citations Over TimeTop 1% of 2017 papers
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 high-order 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.
Related Papers
- → A COMPARATIVE STUDY ON THRESHOLDING TECHNIQUES FOR GRAY IMAGE BINARIZATION(2017)11 cited
- → Multi-level thresholding and its application to feature extraction in machine parts(2005)1 cited
- → Global Thresholding and Multiple Pass Parsing(1997)47 cited
- → A localized thresholding method based on boundary detection(2004)