Presentation | 2021-03-04 What characteristics are acquired in coding self-motion from visual motion? Daiki Nakamura, Hiroaki Gomi, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Efficient coding is a prevailing computational models of sensory coding in the brain. If the sensory information is transformed by this coding mechanism, it could be expected that the neural activities represent sparse and efficient bases decomposed from input signals, independent of the output requirements. While this model was examined as the visual motion coding in the early visual processing, it has not been tested whether or not the characteristics derived by this coding are preserved in the motor output quickly induced by visual motion, which has been considered as a compensatory response to the self-motion. Here we examined this issue by developing a convolutional neural network (CNN) for estimating self-motion from sequential images taken by head-mount camera, and found a spatiotemporal frequency specificity which was similar to that of the motion-induced quick motor response. Interestingly, the specificity was affected by the velocity range of self-motion and a system limitation of the input image sampling. In addition, we analyzed internal representation of the CNN with several indexes employed in describing neural properties in visual cortexes, and found those representation changes in the CNN hierarchy as was found in the brain hierarchical processing. These results suggest that the spatiotemporal frequency specificity of the visually induced motor response is ascribed to the self-motion coding from visual motion rather than the efficient coding. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Visual motion / Self-motion / Convolutional neural network / Manual following response / Ocular following response |
Paper # | NC2020-57 |
Date of Issue | 2021-02-24 (NC) |
Conference Information | |
Committee | NC / MBE |
---|---|
Conference Date | 2021/3/3(3days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Online |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | Neuro Computing, Medical Engineering, etc. |
Chair | Kazuyuki Samejima(Tamagawa Univ) / Takashi Watanabe(Tohoku Univ.) |
Vice Chair | Rieko Osu(Waseda Univ.) / Ryuhei Okuno(Setsunan Univ.) |
Secretary | Rieko Osu(NTT) / Ryuhei Okuno(ATR) |
Assistant | Ken Takiyama(TUAT) / Nobuhiko Wagatsuma(Toho Univ.) / Akihiro Karashima(Tohoku Inst. of Tech.) / Jun Akazawa(Meiji Univ. of Integrative Medicine) |
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) | What characteristics are acquired in coding self-motion from visual motion? |
Sub Title (in English) | Reconstruction of statistical relationship by neural network and its internal representation |
Keyword(1) | Visual motion |
Keyword(2) | Self-motion |
Keyword(3) | Convolutional neural network |
Keyword(4) | Manual following response |
Keyword(5) | Ocular following response |
1st Author's Name | Daiki Nakamura |
1st Author's Affiliation | NTT Communication Science Laboratories(NTT) |
2nd Author's Name | Hiroaki Gomi |
2nd Author's Affiliation | NTT Communication Science Laboratories(NTT) |
Date | 2021-03-04 |
Paper # | NC2020-57 |
Volume (vol) | vol.120 |
Number (no) | NC-403 |
Page | pp.pp.83-88(NC), |
#Pages | 6 |
Date of Issue | 2021-02-24 (NC) |