Presentation | 2018-12-15 Spatial frequency characteristics of convolutional neutral network trained for classifying facial expressions Yusuke Komatsu, Mikio Inagaki, ChanSeok Lim, Takashi Shinozaki, Ichiro Fujita, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | A previous experiment conducted in monkeys (Inagaki and Fujita, 2011) demonstrated that most face-responsive neurons in the temporal visual cortex are tuned to image-based spatial frequency (SF) (cycles/image), not to retina-based SF (cycles/degree). In this study, we addressed how the tuning to image-based SF observed in face-responsive neurons contributes to underlying computation of face classification. We analyzed the internal representation of a convolutional neural network (CNN) which imitates processing of the ventral visual pathway. We found a positive correlation between the preference for image-based SF and classification performance, suggesting that this tuning property plays a role in discriminating facial expressions. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Face recognition / Face-responsive neuron / Convolutional Neural Network / Spatial Frequency |
Paper # | NC2018-29 |
Date of Issue | 2018-12-08 (NC) |
Conference Information | |
Committee | NC / MBE |
---|---|
Conference Date | 2018/12/15(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Nagoya Institute of Technology |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Yutaka Hirata(Chubu Univ.) / Masaki Kyoso(TCU) |
Vice Chair | Hayaru Shouno(UEC) / Taishin Nomura(Osaka Univ.) |
Secretary | Hayaru Shouno(Nagoya Univ.) / Taishin Nomura(NAIST) |
Assistant | Keiichiro Inagaki(Chubu Univ.) / Takashi Shinozaki(NICT) / Takumi Kobayashi(YNU) / Yasuyuki Suzuki(Osaka Univ.) |
Paper Information | |
Registration To | Technical Committee on Neurocomputing / Technical Committee on ME and Bio Cybernetics |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Spatial frequency characteristics of convolutional neutral network trained for classifying facial expressions |
Sub Title (in English) | |
Keyword(1) | Face recognition |
Keyword(2) | Face-responsive neuron |
Keyword(3) | Convolutional Neural Network |
Keyword(4) | Spatial Frequency |
1st Author's Name | Yusuke Komatsu |
1st Author's Affiliation | Osaka University(Osaka Univ) |
2nd Author's Name | Mikio Inagaki |
2nd Author's Affiliation | Osaka University(Osaka Univ) |
3rd Author's Name | ChanSeok Lim |
3rd Author's Affiliation | Osaka University(Osaka Univ) |
4th Author's Name | Takashi Shinozaki |
4th Author's Affiliation | National Institute of information and Communications Technology(NICT) |
5th Author's Name | Ichiro Fujita |
5th Author's Affiliation | Osaka University(Osaka Univ) |
Date | 2018-12-15 |
Paper # | NC2018-29 |
Volume (vol) | vol.118 |
Number (no) | NC-367 |
Page | pp.pp.5-10(NC), |
#Pages | 6 |
Date of Issue | 2018-12-08 (NC) |