Download presentation
Presentation is loading. Please wait.
Published byCory Tucker Modified over 8 years ago
1
Compression of CNNs Mooyeol Baek Xiangyu Zhang, Jianhua Zou, Xiang Ming, Kaiming He, Jian Sun: Efficient and Accurate Approximations of Nonlinear Convolutional Networks. Yong-Deok Kim, Eunhyeok Park, Sungjoo Yoo, Taelim Choi, Lu Yang, Dongjun Shin: Compression of Deep Convolutional Neural Networks for Fast and Low Power Mobile Applications.
2
Motivation It’s practically important to accelerate the test-time computation of CNNs. CNN filters can be approximately decomposed into a series of smaller filters by row-rank approximation.
3
Approaches Zhang et al. Kim et al. m m c n d k k c m m c 1 1 c m m c’ n n d’ 1 1 n d n n 1 1 k k c’ n n
4
Efficient and Accurate Approximations of Nonlinear Convolutional Networks. Xiangyu Zhang, Jianhua Zou, Xiang Ming, Kaiming He, Jian Sun
5
Contribution Low-rank approximation minimizing the reconstruction error of nonlinear responses. Asymmetric reconstruction to reduce the accumulated error of multiple approximated layers. Empirical observation of PCA energy to select proper rank.
6
Low-rank Approximation m m c n d k k c n n d’ 1 1 n
7
Low-rank Approximation Relaxation
8
Asymmetric Reconstruction Uses non-approximate responses to reduce the accumulated error of multiple approximated layers. OriginalApproximated
9
Rank Selection
10
Experiments [1] Linear vs. Nonlinear
11
Experiments [2] Symmetric vs. Asymmetric
12
Experiments [3] Rank selection
13
Compression of Deep Convolutional Neural Networks for Fast and Low Power Mobile Applications. Yong-Deok Kim, Eunhyeok Park, Sungjoo Yoo, Taelim Choi, Lu Yang, Dongjun Shin
14
Contribution One-shot whole network compression scheme which consists of simple three steps: 1.Rank selection (Variational Bayesian matrix factorization) 2.Low-rank tensor decomposition (Tucker decomposition) 3.Fine-tuning.
15
Tensor Decomposition Tucker decomposition
16
Tensor Decomposition Zhang et al. Kim et al. m m c n d k k c m m c 1 1 c m m c’ n n d’ 1 1 n d n n 1 1 k k c’ n n
17
Fine-tuning
18
Experiments [1]
19
Experiments [2]
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.