Exploiting Linear Structure Within Convolutional Networks for Efficient Evaluation
Emily Denton, Wojciech Zaremba, Joan Bruna et al. · arXiv (Cornell University) · 2014
We present techniques for speeding up the test-time evaluation of large convolutional networks, designed for object recognition tasks. These models deliver impressive accuracy but each image evaluation requires millions of floating point operations, making their deployment on smartphones and Internet-scale clusters problematic. The computation is dominated by the convolution operations in the lower layers of the model. We exploit the linear structure present within the convolutional filters to derive approximations that significantly reduce the required computation. Using large state-of-the-art models, we demonstrate we demonstrate speedups of convolutional layers on both CPU and GPU by a factor of 2x, while keeping the accuracy within 1% of the original model.
Cited by 1,300 works. doi:10.48550/arxiv.1404.0736 Open-access PDF
Authors
References
- ImageNet classification with deep convolutional neural networks
- ImageNet: A large-scale hierarchical image database
- Improving neural networks by preventing co-adaptation of feature detectors
- Object recognition from local scale-invariant features
- OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks
- Visualizing and Understanding Convolutional Neural Networks
- Adaptive deconvolutional networks for mid and high level feature learning
- Improving the speed of neural networks on CPUs
- Tiled convolutional neural networks
- Building high-level features using large scale unsupervised learning