Learning Filter Basis for Convolutional Neural Network Compression (2019-08-23T00:00:00.000000Z)

TL;DR

This paper tries to reduce the number of parameters of CNNs by learning a basis of the filters in convolutional layers, and validate the proposed solution for multiple CNN architectures on image classification and image super-resolution benchmarks.

Abstract

Convolutional neural networks (CNNs) based solutions have achieved state-of-the-art performances for many computer vision tasks, including classification and super-resolution of images. Usually the success of these methods comes with a cost of millions of parameters due to stacking deep convolutional layers. Moreover, quite a large number of filters are also used for a single convolutional layer, which exaggerates the parameter burden of current methods. Thus, in this paper, we try to reduce the number of parameters of CNNs by learning a basis of the filters in convolutional layers. For the forward pass, the learned basis is used to approximate the original filters and then used as parameters for the convolutional layers. We validate our proposed solution for multiple CNN architectures on image classification and image super-resolution benchmarks and compare favorably to the existing state-of-the-art in terms of reduction of parameters and preservation of accuracy. Code is available at https://github.com/ofsoundof/learning_filter_basis

Authors

L. Gool

68 papers

Shuhang Gu

2 papers

Radu Timofte

49 papers

TL;DR

Abstract

Authors

References55 items

3D Appearance Super-Resolution With Deep Learning

Deep Plug-And-Play Super-Resolution for Arbitrary Blur Kernels

Cascaded Projection: End-To-End Network Compression and Acceleration

Exploiting Kernel Sparsity and Entropy for Interpretable CNN Compression

Efficient Neural Network Compression

Clustering Convolutional Kernels to Compress Deep Neural Networks

CARN: Convolutional Anchored Regression Network for Fast and Accurate Single Image Super-Resolution

Multi-bin Trainable Linear Unit for Fast Image Restoration Networks

Extreme Network Compression via Filter Group Approximation

AMC: AutoML for Model Compression and Acceleration on Mobile Devices

Joint Blind Motion Deblurring and Depth Estimation of Light Field

Automatic differentiation in PyTorch

Learning Efficient Convolutional Networks through Network Slimming

NTIRE 2017 Challenge on Single Image Super-Resolution: Dataset and Study

Channel Pruning for Accelerating Very Deep Neural Networks

Enhanced Deep Residual Networks for Single Image Super-Resolution

Learning Deep CNN Denoiser Prior for Image Restoration

Incremental Network Quantization: Towards Lossless CNNs with Low-Precision Weights

ECO: Efficient Convolution Operators for Tracking

Trained Ternary Quantization

Learning the Number of Neurons in Deep Networks

Less Is More: Towards Compact CNNs

Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network

Densely Connected Convolutional Networks

Factorized Convolutional Neural Networks

Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising

Learning Structured Sparsity in Deep Neural Networks

Blind Image Deblurring Using Dark Channel Prior

XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks

Binarized Neural Networks

Deep Residual Learning for Image Recognition

Accurate Image Super-Resolution Using Very Deep Convolutional Networks

BinaryConnect: Training Deep Neural Networks with binary weights during propagations

Deep Compression: Compressing Deep Neural Network with Pruning, Trained Quantization and Huffman Coding

You Only Look Once: Unified, Real-Time Object Detection

Learning both Weights and Connections for Efficient Neural Network

Single image super-resolution from transformed self-exemplars

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Accelerating Very Deep Convolutional Networks for Classification and Detection

Fast R-CNN

Compressing Neural Networks with the Hashing Trick

Adam: A Method for Stochastic Optimization

Speeding-up Convolutional Neural Networks Using Fine-tuned CP-Decomposition

Fully convolutional networks for semantic segmentation

Learning a Deep Convolutional Network for Image Super-Resolution

Very Deep Convolutional Networks for Large-Scale Image Recognition

Speeding up Convolutional Neural Networks with Low Rank Expansions

Exploiting Linear Structure Within Convolutional Networks for Efficient Evaluation

Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation

ImageNet classification with deep convolutional neural networks

Low-Complexity Single-Image Super-Resolution based on Nonnegative Neighbor Embedding

On Single Image Scale-Up Using Sparse-Representations

A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics

Learning Multiple Layers of Features from Tiny Images

This Paper Is Included in the Proceedings of the 12th Usenix Symposium on Operating Systems Design and Implementation (osdi '16). Tensorflow: a System for Large-scale Machine Learning Tensorflow: a System for Large-scale Machine Learning

Field of Study

Journal Information

Name

Page

Venue Information

Name

Type

URL

Alternate Names