LCNN: Lookup-Based Convolutional Neural Network

Hessam Bagherinezhad, Mohammad Rastegari, Ali Farhadi; The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 7120-7129

Abstract


Porting state of the art deep learning algorithms to resource constrained compute platforms (e.g. VR, AR, wearables) is extremely challenging. We propose a fast, compact, and accurate model for convolutional neural networks that enables efficient learning and inference. We introduce LCNN, a lookup-based convolutional neural network that encodes convolutions by few lookups to a dictionary that is trained to cover the space of weights in CNNs. Training LCNN involves jointly learning a dictionary and a small set of linear combinations. The size of the dictionary naturally traces a spectrum of trade-offs between efficiency and accuracy. Our experimental results on ImageNet challenge show that LCNN can offer 3.2x speedup while achieving 55.1% top-1 accuracy using AlexNet architecture. Our fastest LCNN offers 37.6x speed up over AlexNet while maintaining 44.3% top-1 accuracy. LCNN not only offers dramatic speed ups at inference, but it also enables efficient training. In this paper, we show the benefits of LCNN in few-shot learning and few-iteration learning, two crucial aspects of on-device training of deep learning models.

Related Material


[pdf] [Supp] [arXiv]
[bibtex]
@InProceedings{Bagherinezhad_2017_CVPR,
author = {Bagherinezhad, Hessam and Rastegari, Mohammad and Farhadi, Ali},
title = {LCNN: Lookup-Based Convolutional Neural Network},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {July},
year = {2017}
}