Aim and Topics

Various textures perceived through the five senses are used in our decision making and executions. Our level of understanding and methodology differ in each of the five senses. Researches in visual-texture perception are said to be the most developed among all. However, the neural mechanism of visual-texture perception remains unclear for the most part. Thus, there is no established way to reproduce the mechanism with artificial intelligence.

In the field of tactile-texture perception research, where tactile has been considered to be primitive compared to vision, technologies to display tactile sensation using artificial intelligent are awaited. Various of problems are making them difficult to be realized: Auditory-texture perception caused by complex signal modulation, Taste-texture perception inseparable from visual sense and sense of smell, and elucidation of the relationship of higher-order sensitivity processing in brain. Approaches based on statistical machine learning represented by ‘deep-learning’ which has been remarkably developed in recent years are being applied to understand the overview of the brain that constitutes the texture perception network. These attempts are mainly focused on visual-texture aspects and few are reported related to tactile-texture so far.

This WS is to be an opportunity to share the progress of research and to share methodology by researchers studying the five senses. The scope of this WS shall be: (1) research of science and engineering related to value judgements made through the five senses, such as image processing, tactile engineering, acoustics, machine learning, sensitivity engineering, and natural language processing (2) research of biological systems, psychophysical, and neuroscience. We would like to discuss the scientific understanding of texture information processing and the possibility of developing new artificial intelligence research.

Registration

Please register the workshop at registration page of JSAI International Symposia on AI 2020.

Program

November 17 (Tue)

11:20-11:30: Opening remark

11:30-12:00: Style Image Retrieval for Improving Material Translation Using Neural Style Transfer, Gibran Benitez-Garcia, Wataru Shimoda, and Keiji Yanai

12:00-13:00: Lunch

13:00-13:30: Onomatopoeia generation system expressing texture of image, Koichi YAMAGATA, Jinhwan KWON, Takuya KAWASHIMA, Wataru SHIMODA, and Maki SAKAMOTO

13:30-13:40: Closing

Contact

koichi.yamagata[at]uec.ac.jp