%0 Generic %1 wen2016learning %A Wen, Wei %A Wu, Chunpeng %A Wang, Yandan %A Chen, Yiran %A Li, Hai %D 2016 %K deep_learning lasso learning performance regularization sparse sparsity %T Learning Structured Sparsity in Deep Neural Networks %U http://arxiv.org/abs/1608.03665 %X High demand for computation resources severely hinders deployment of large-scale Deep Neural Networks (DNN) in resource constrained devices. In this work, we propose a Structured Sparsity Learning (SSL) method to regularize the structures (i.e., filters, channels, filter shapes, and layer depth) of DNNs. SSL can: (1) learn a compact structure from a bigger DNN to reduce computation cost; (2) obtain a hardware-friendly structured sparsity of DNN to efficiently accelerate the DNNs evaluation. Experimental results show that SSL achieves on average 5.1x and 3.1x speedups of convolutional layer computation of AlexNet against CPU and GPU, respectively, with off-the-shelf libraries. These speedups are about twice speedups of non-structured sparsity; (3) regularize the DNN structure to improve classification accuracy. The results show that for CIFAR-10, regularization on layer depth can reduce 20 layers of a Deep Residual Network (ResNet) to 18 layers while improve the accuracy from 91.25% to 92.60%, which is still slightly higher than that of original ResNet with 32 layers. For AlexNet, structure regularization by SSL also reduces the error by around ~1%. Open source code is in https://github.com/wenwei202/caffe/tree/scnn

@misc{wen2016learning, abstract = {High demand for computation resources severely hinders deployment of large-scale Deep Neural Networks (DNN) in resource constrained devices. In this work, we propose a Structured Sparsity Learning (SSL) method to regularize the structures (i.e., filters, channels, filter shapes, and layer depth) of DNNs. SSL can: (1) learn a compact structure from a bigger DNN to reduce computation cost; (2) obtain a hardware-friendly structured sparsity of DNN to efficiently accelerate the DNNs evaluation. Experimental results show that SSL achieves on average 5.1x and 3.1x speedups of convolutional layer computation of AlexNet against CPU and GPU, respectively, with off-the-shelf libraries. These speedups are about twice speedups of non-structured sparsity; (3) regularize the DNN structure to improve classification accuracy. The results show that for CIFAR-10, regularization on layer depth can reduce 20 layers of a Deep Residual Network (ResNet) to 18 layers while improve the accuracy from 91.25% to 92.60%, which is still slightly higher than that of original ResNet with 32 layers. For AlexNet, structure regularization by SSL also reduces the error by around ~1%. Open source code is in https://github.com/wenwei202/caffe/tree/scnn}, added-at = {2018-06-13T17:29:10.000+0200}, author = {Wen, Wei and Wu, Chunpeng and Wang, Yandan and Chen, Yiran and Li, Hai}, biburl = {https://www.bibsonomy.org/bibtex/208a55e1f60d783dca875d5fe205d61bf/loroch}, description = {[1608.03665] Learning Structured Sparsity in Deep Neural Networks}, interhash = {3ee1eaaaff07cb3079459e838df37c83}, intrahash = {08a55e1f60d783dca875d5fe205d61bf}, keywords = {deep_learning lasso learning performance regularization sparse sparsity}, note = {cite arxiv:1608.03665Comment: Accepted by NIPS 2016}, timestamp = {2018-06-16T10:46:03.000+0200}, title = {Learning Structured Sparsity in Deep Neural Networks}, url = {http://arxiv.org/abs/1608.03665}, year = 2016 }