Presentation | 2020-03-17 Rule Extraction from convolutional neural networks Yuya Sato, Hiroshi Tsukimoto, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | We presented a decompositional method of rule extraction from convolutional neural networks in the past. However, we found that the accuracy was extremely low in the case of fully connected layers with hidden layers. This paper presents a pedagogical extracting method. Also, this paper presents a method merging rules and images maximizing the output of convolutional/pooling layers, and visualizing features of the rules extracted from fully connected layers. These methods were applied to convolutional neural networks trained with MNIST dataset. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | convolutional neural network / feature extraction / rule extraction / SmoothGrad / MNIST / pedagogical method |
Paper # | PRMU2019-89 |
Date of Issue | 2020-03-09 (PRMU) |
Conference Information | |
Committee | PRMU / IPSJ-CVIM |
---|---|
Conference Date | 2020/3/16(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Yoichi Sato(Univ. of Tokyo) |
Vice Chair | Toru Tamaki(Hiroshima Univ.) / Akisato Kimura(NTT) |
Secretary | Toru Tamaki(NTT) / Akisato Kimura(OMRON SINICX) |
Assistant | Yusuke Uchida(DeNA) / Takayoshi Yamashita(Chubu Univ.) |
Paper Information | |
Registration To | Technical Committee on Pattern Recognition and Media Understanding / Special Interest Group on Computer Vision and Image Media |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Rule Extraction from convolutional neural networks |
Sub Title (in English) | Rule Extraction with A Pedagogical Method |
Keyword(1) | convolutional neural network |
Keyword(2) | feature extraction |
Keyword(3) | rule extraction |
Keyword(4) | SmoothGrad |
Keyword(5) | MNIST |
Keyword(6) | pedagogical method |
1st Author's Name | Yuya Sato |
1st Author's Affiliation | Tokyo Denki University(Tokyo Denki Univ) |
2nd Author's Name | Hiroshi Tsukimoto |
2nd Author's Affiliation | Tokyo Denki University(Tokyo Denki Univ.) |
Date | 2020-03-17 |
Paper # | PRMU2019-89 |
Volume (vol) | vol.119 |
Number (no) | PRMU-481 |
Page | pp.pp.121-126(PRMU), |
#Pages | 6 |
Date of Issue | 2020-03-09 (PRMU) |