模型通道剪枝汇总（channel pruning）

模型通道剪枝总结

Pruning Filters for Efficient ConvNets
论文：https://arxiv.org/pdf/1608.08710.pdf
github：https://github.com/666DZY666/model-compression
解读：https://zhuanlan.zhihu.com/p/98365407

ThiNet: A Filter Level Pruning Method for Deep Neural Network Compression
论文：https://arxiv.org/pdf/1707.06342.pdf
github：https://github.com/Roll920/ThiNet
解读：https://blog.csdn.net/liujianlin01/article/details/80738741

Channel Pruning for Accelerating Very Deep Neural Networks
论文：https://arxiv.org/pdf/1707.06168.pdf
github：https://github.com/yihui-he/channel-pruning、
https://github.com/Tencent/PocketFlow
解读：https://blog.csdn.net/u014380165/article/details/79811779、https://zhuanlan.zhihu.com/p/87791509
PocketFlow使用文档：https://pocketflow.github.io/cp_learner/

Learning Efficient Convolutional Networks Through Network Slimming
论文：https://openaccess.thecvf.com/content_iccv_2017/html/Liu_Learning_Efficient_Convolutional_ICCV_2017_paper.html
解读：https://blog.csdn.net/h__ang/article/details/89376079

关于这篇论文和相关的yolov3项目，我做了一个总结，传送门在这里,有完整代码，复现很方便：https://blog.csdn.net/qq_38109843/article/details/107234801

Data-Driven Sparse Structure Selection for Deep Neural Networks
论文：https://arxiv.org/pdf/1707.01213.pdf
github：https://github.com/TuSimple/sparse-structure-selection、https://github.com/Pokemon-Huang/sparse-structure-selection-PyTorch
解读：https://blog.csdn.net/sinat_35188997/article/details/78180767

Soft Filter Pruning for Accelerating Deep Convolutional Neural Networks
论文：https://www.ijcai.org/Proceedings/2018/0309.pdf
github：https://github.com/he-y/soft-filter-pruning
解读：https://blog.csdn.net/u014380165/article/details/81107032

Rethinking the Value of Network Pruning
论文：https://arxiv.org/pdf/1810.05270.pdf
github：https://github.com/Eric-mingjie/rethinking-network-pruning
解读：https://www.jiqizhixin.com/articles/2018-10-22-6

Rethinking the Smaller-Norm-Less-Informative Assumption in Channel Pruning of Convolution Layers
论文：https://arxiv.org/abs/1802.00124
github：https://github.com/bobye/batchnorm_prune
解读：https://blog.csdn.net/hw5226349/article/details/84779325

MorphNet: Fast & Simple Resource-Constrained Structure Learning of Deep
论文：https://arxiv.org/pdf/1711.06798.pdf
github：https://github.com/google-research/morph-net
解读：https://zhuanlan.zhihu.com/p/62888707

Slimmable Networks三部曲
github：https://github.com/JiahuiYu/slimmable_networks
解读：https://www.zhihu.com/question/306865592
总结：https://zhuanlan.zhihu.com/p/105064255
SLIMMABLE NEURAL NETWORKS
论文：https://arxiv.org/pdf/1812.08928.pdf
解读：https://blog.csdn.net/qq_14845119/article/details/89453059
Universally Slimmable Networks and Improved Training Techniques
论文：https://arxiv.org/pdf/1903.05134.pdf
解读：https://blog.csdn.net/lihuanyu520/article/details/107482145
AutoSlim: Towards One-Shot Architecture Search for Channel Numbers
论文：https://arxiv.org/pdf/1903.11728.pdf
解读：https://zhuanlan.zhihu.com/p/105064255

Discrimination-aware Channel Pruning for Deep Neural Networks
论文：https://arxiv.org/pdf/1810.11809.pdf
github：https://github.com/Tencent/PocketFlow
解读：https://blog.csdn.net/nature553863/article/details/83822895
PocketFlow使用文档：https://pocketflow.github.io/dcp_learner/

AMC: AutoML for Model Compression and Acceleration on Mobile Devices
论文：https://arxiv.org/pdf/1802.03494.pdf
github:https://github.com/mit-han-lab/amc
解读：https://blog.csdn.net/librahfacebook/article/details/100049955
tips：https://blog.csdn.net/qq_30615903/article/details/102588437

Filter Pruning via Geometric Median for Deep Convolutional Neural Network Acceleration
论文：https://arxiv.org/pdf/1811.00250.pdf
github：https://github.com/he-y/filter-pruning-geometric-median
解读：https://blog.csdn.net/qq_37643960/article/details/101228726、https://zhuanlan.zhihu.com/p/67175001

Channel Pruning via Optimal Thresholding
论文：https://arxiv.org/pdf/2003.04566.pdf
github：https://github.com/yeyun11/netslim
解读：https://blog.csdn.net/c2250645962/article/details/104921610/、https://zhuanlan.zhihu.com/p/74553341

开源工具：
Inter
github：https://github.com/NervanaSystems/distiller
使用文档：https://nervanasystems.github.io/distiller/usage.html

PocketFlow
github：https://github.com/Tencent/PocketFlow
使用文档：https://pocketflow.github.io/tutorial/

nni
github：https://github.com/Microsoft/nni/
使用文档：https://nni.readthedocs.io/zh/latest/Compressor/Overview.html

PaddleSlim
github：https://github.com/PaddlePaddle/PaddleSlim
使用文档：https://paddlepaddle.github.io/PaddleSlim/