| Presentation | 2003/1/27 Neural network model for search asymmetry in visual search task Takahiko KOIKE, Jun SAIKI, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Using visual attention, our visual system selects important information in our environment. Some models dealt with the mechanism of attentional shift, and they could well simulate the characteristics of human performance in simple visual search tasks, e.g. pop-out search and conjunction search. However, these earlier models do not seem to be able to explain why search asymmetry in visual search occurs. Search asymmetry is likely to reflect changes in relative saliency between a target and distractors, but the earlier models with a deterministic WTA ignore relative saliency because they always shift the focus of attention solely based on the rank-order of the saliency. We proposed a stochastic saliency model, and showed that the introduction of a stochastic WTA enables the saliency-based search model to cause changes in search efficiency through stochastic shifts of attentional focus. Our proposed model also well simulates the triple-conjunction search task. This result suggests that our visual system may achieve high-speed information processing in exchange for rough and stochastic processing. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Visual attention / Visual search task / Search asymmetry / Saliency map / Stochastic WTA |
| Paper # | NC2002-107 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 2003/1/27(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Neurocomputing (NC) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Neural network model for search asymmetry in visual search task |
| Sub Title (in English) | |
| Keyword(1) | Visual attention |
| Keyword(2) | Visual search task |
| Keyword(3) | Search asymmetry |
| Keyword(4) | Saliency map |
| Keyword(5) | Stochastic WTA |
| 1st Author's Name | Takahiko KOIKE |
| 1st Author's Affiliation | Department of Intelligence Science and Technology, Graduate School of Informatics, Kyoto University() |
| 2nd Author's Name | Jun SAIKI |
| 2nd Author's Affiliation | Japan Science and Technology Corporation (JST):Department of Intelligence Science and Technology, Graduate School of Informatics, Kyoto University |
| Date | 2003/1/27 |
| Paper # | NC2002-107 |
| Volume (vol) | vol.102 |
| Number (no) | 627 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |