Kyushu University
Faculty of Design

Profile

He is interested in Computational Intelligence such as NN, FS, GA, and other so-called Soft Computing technologies, especially cooperation of these technologies. Currently, his interest focuses on Interactive Evolutionary Computation which aims the cooperation of human and evolutionary computation. (more ....)

Humanized Computational Intelligence

In the latter half of 1980s, neural networks, fuzzy systems, and evolutionary computation became fundamental engineering technologies and have widely
used in many fields practically. Their cooperative techniques have been used as well. See detail.

What comes next? I believe that systems where a human and computational intelligence cooperate and maximize human capability for the computational intelligence will become important. For example, it is impossible to realize cute movement of a robot and virtual reality control to let a human feel as if the user is fitting with a real human. without incorporating human evaluation into their engineering designs; we cannot handle the quality of knowledge acquired using data mining techniques without the evaluation of a domain expert; there are many signal processing applications that human evaluate the signal processing quality though visual inspection or auditory inspection such as hearing-aid fitting or medical image enhancement..

I believe that Humanized Computational Intelligence must become one of important research directions. Interactive evolutionary computation is one of technical tools that realize this research direction. First four pages of the slide show this my view in detail..

Interactive Evolutionary Computation

Interactive Evolutionary Computation (IEC) is a technique that evolutionary computation consisting of genetic algorithms (GA), evolutionary strategy (ES), evolutionary programming (EP), and genetic programming (GP) optimizes the target system based human subjective evaluation.. When specifications or design goal of the target system is numerically given, normal optimization techniques can be used. However, there are many cases that the system performance is not measurable and only human can evaluate the system performance, for example, maximizing sound quality of a hearing aid for a user, generating computer graphics for my living room, generating Jazz-like music. Subjective evaluation includes both KANSEI scale such as preference and evaluation based on domain knowledge.

We need a technical framework for combining human evaluation and computer optimization for such tasks. IEC is one of techniques that realize this framework.

IEC is widely used in artistic field, engineering field, and other fields. User interface research to reduce IEC user's fatigue is another research topic. If you are interested in IEC, the best introduction is my tutorial paper that includes survey of 251 IEC papers. My slides that introduces many IEC applications is another tutorial material. There are links to tutorial figures and short descriptions from research topics at my laboratory web page.