.png){kind=icon}
返回
AutoPruner: An end-to-end trainable filter pruning method for efficient deep model inference
.png){kind=icon}
OA
AI
DOI：10.1016/j.patcog.2020.107461.png)
.png){kind=icon}
.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Channel pruning is an important method to speed up CNN model's inference. Previous filter pruning algorithms regard importance evaluation and model fine-tuning as two independent steps. This paper argues that combining them into a single end-to-end trainable system will lead to better results. We propose an efficient channel selection layer, namely AutoPruner, to find less important filters automatically in a joint training manner. Our AutoPruner takes previous activation responses as an input and generates a true binary index code for pruning. Hence, all the filters corresponding to zero index values can be removed safely after training. By gradually erasing several unimportant filters, we can prevent an excessive drop in model accuracy. Compared with previous state-of-the-art pruning algorithms (including training from scratch), AutoPruner achieves significantly better performance. Furthermore, ablation experiments show that the proposed novel mini-batch pooling and binarization operations are vital for the success of model pruning. (C) 2020 Elsevier Ltd. All rights reserved.
Keyword:
Neural network pruning
Model compression
CNN acceleration
AI总结
对已上传原文的论文进行重点信息的提取，主要内容包括：简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。
期刊
IF:
7.6
论文数:
1.3W
被引数:
4.5W
机构
引用论文
LightweightNet: Toward fast and lightweight convolutional neural networks via architecture distillation.png)
PATTERN RECOGNITION
PATTERN RECOGNITIONIF7.6
PREAI
分享收藏Building fast and compact convolutional neural networks for offline handwritten Chinese character recognitionPATTERN RECOGNITIONOAAI
PATTERN RECOGNITIONIF7.6
OAAI分享收藏Gradient-based learning applied to document recognition基于梯度的学习在文档识别中的应用PROCEEDINGS OF THE IEEE
PROCEEDINGS OF THE IEEEIF25.9
PREAI
分享收藏Toward Compact ConvNets via Structure-Sparsity Regularized Filter Pruning通过结构稀疏正则化滤波器修剪实现紧凑的ConvNets
PREAI
分享收藏分享收藏 没有更多内容
没有更多内容