Dear my friends in the world,
I made this 2007 annual report as my seasonal greeting to my friends.
Please enjoy it. I hope to keep in touch with you.
Thinking of you from Japan in early 2007
--- My topics in 2006 ---
During the 2006 fiscal year (April, 2006 - March, 2007), I have 3 Bachelor students, 6 Master students, 2 Doctoral students, and maximum 3 postdoctoral researchers. In the 2007 fiscal year, I will have about 2 Bachelor students, 4 Master students, and 2 doctoral students. Unfortunately, there will be no positions for postdoctoral researchers in 2007.
My student, Nobuhisa Tanaka, who finished a doctoral course at the end of March, 2006, completed his doctoral dissertation entitled A Study on the Design of the Virtual Reality Environment Taking into Account Individual User Characteristics and will receive a doctoral degree in January, 2007. As the characteristics of VR sickness and VR perception depend on persons, it is difficult to design a VR environment that minimizes VR sickness of a user and maximizes his/her VR perception. He introduced neural networks (NN's) to learn these two personal characteristics previously with short pre-measurements, made the NN's estimate the best VR conditions for each VR user, and evaluated his proposed VR environment design system.
I have worked on the combination of human factors and computational intelligence that I call as Humanized Computational Intelligence. Interactive evolutionary computation (IEC) is still a main technique
in my research for this direction. I prepare a survey and tutorial paper
of IEC research and a PDF file of slides of IEC applications at the top
of my web page, FYI.
There are three funded projects that I am focusing:
All of them are research on designing artificial environment and are deeply related each other.
Simple explanation of "Control of multi-media users' emotion based on IEC and physiological analysis" would be "to make it clear how to realize more impressive, more calm, and others." When we view movies or listen music, sometimes we become impressive, relax, thrilling, or depressive. These emotional responses are reflected on physiological reactions. The most strong factor giving influence on such emotion responses may be context, e.g. stories of movies, but in the sensory level, physical signals must influence physiological responses directly. This research aims to find out physical features that influence physiological responses, find out their relationships, and control the physical features to make human's physiological responses move to the target responses. As the first step, in 2005, we extracted several physical movie features that may influence on our physiological reactions; as the second step, in 2006, we made subjects watch abstract movies and measured their EEG and other physiological signals. Now we are analyzing the relationship between extracted movie physical features and measured physiological data; as the third step, in 2007, we are going to control movie physical features directly and control human physiological reaction indirectly if we find out definite relationship between two. I am thinking to use Extended IEC based on physiological feedback for the indirect control.
Our current approach for a research on Reduction of IEC User's Fatigue is to embed a model learning the evaluation characteristics of an IEC user in an IEC system, estimate the subjective evaluation value of the user, and accelerate IEC optimization using the estimated evaluation values, i.e. reducing the number of IEC user's evaluation by learning the user's evaluation characteristics. IEC has several restrictions that normal optimizations do not have, for example, it is hard to obtain enough number of training data because an IEC user cannot evaluate individuals many times, and the generations in which a trained learning model works are not many because an IEC user cannot continue to evaluate for many generations.
We are trying to solve these difficulties by (1) embedding multiple evaluation models of other several users that are learnt previously into an IEC system and using the models as alternative of the IEC user until the system leans out the user's evaluation characteristics, (2) adopting several learning models and finding out better models for learning IEC user's evaluation characteristics, and (3) evaluating interactive particle swarm optimization (PSO) instead of IEC.
Two students started doctoral researches since April, 2006.
Doctoral research on Modeling Naturalness in Motion is to model as "human motions = global motions modeled mathematically + natural motions", make a recurrent neural network (RNN) learn the difference between measured human motions and the former motions generated by mathematical models, and generate natural motions in robot motions or computer graphics. As the first step, he is modeling static hand-written letters with "fonts + personal fluctuation" instead of dynamic motions, and making a RNN learn the latter, and trying to generating natural hand-written letters.
Doctoral research on Architectural Design Support Using Multi-objective Optimization and IEC is being conducted using application tasks of designing layouts of rooms, buildings/houses within given land shape, or garden. These tasks have multi-objectives under several design restrictions and include several design factors that it is hard to evaluate quantitatively. He is trying to solve these tasks by combining evolutionary multi-objective optimization and IEC.
Our research outputs in 2006 was 3 published and accepted journal papers, 5 international conference papers, and 5 domestic conference papers.
I taught courses on computational intelligence, software engineering, data and signal compression, academic English A, and other lab works in 2006.
Major Activities in 2006