Rapid Exact Signal Scanning With Deep Convolutional Neural Networks

2015/08/31 by Markus Thom, Franz Gritschneder
Computer Science · #Algorithm #Artificial intelligence #Artificial neural network #Blind Source Separation Techniques #Computation #Computational complexity theory #Computer engineering #Computer hardware #Computer science #Convolutional neural network #Digital signal processing #Image and Signal Denoising Methods #Massively parallel #Neural Networks and Applications #Parallel computing #SIGNAL (programming language) #Signal processing #cs.CV #cs.LG #cs.NE

paper · pdf · doi:10.1109/tsp.2016.2631454

published as IEEE Transactions on Signal Processing, vol. 65, no. 5, pp. 1235-1250 (2017) · Pages 1-16 only: Copyright (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission

openalex publication_date 2016/11/23 · arxiv created 2017/08/02 · arxiv updated 2017/08/03 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/05

Abstract

A rigorous formulation of the dynamics of a signal processing scheme aimed at dense signal scanning without any loss in accuracy is introduced and analyzed. Related methods proposed in the recent past lack a satisfactory analysis of whether they actually fulfill any exactness constraints. This is improved through an exact characterization of the requirements for a sound sliding window approach. The tools developed in this paper are especially beneficial if Convolutional Neural Networks are employed, but can also be used as a more general framework to validate related approaches to signal scanning. The proposed theory helps to eliminate redundant computations and renders special case treatment unnecessary, resulting in a dramatic boost in efficiency particularly on massively parallel processors. This is demonstrated both theoretically in a computational complexity analysis and empirically on modern parallel processors.

Citations

• Highly Efficient Forward and Backward Propagation of Convolutional Neural Networks for Pixelwise Classification
• Learning representations by back-propagating errors
• DenseNet: Implementing Efficient ConvNet Descriptor Pyramids
• cuDNN: Efficient Primitives for Deep Learning
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Image Super-Resolution Using Deep Convolutional Networks
• Gradient-based learning applied to document recognition
• ImageNet Large Scale Visual Recognition Challenge
• Multi-column deep neural networks for image classification
• ImageNet classification with deep convolutional neural networks
• DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs
• ZNNi - Maximizing the Inference Throughput of 3D Convolutional Networks on Multi-Core CPUs and GPUs
• DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs
• Fully Convolutional Networks for Semantic Segmentation