Masaaki Ida

Behavior-based fuzzy control system for a mobile robot with environment recognition by sensory-motor coordination

An idea to form symbolic representations which can be coupled with behavior-based robot systems is introduced. Our behavior-based control system is implemented with Subsumption Architecture (SA), which is a class of behavior-based systems, and the modules of SA are realized with fuzzy control. Our fundamental idea in applying a symbolic environmental model to this behavior-based system is that the categories attached to the symbols are to be extracted from the sensory-motor coordination of the robot. We realized this idea by the use of self-organizing map (SOM) which categorizes behavior sequence of the robot. By categorizing the resulting behaviors, the concept about the categories of the environmental structures, which is embedded in the sensory-motor coordination, can be extracted. In the last part of this paper, we introduce a hybrid system which is composed of a lower-level behavior-based system and a higher-level symbol system, and then is applied to a path-planning problem as a verification of our proposed idea.

Proposal and evaluation of the active course classification support system with exploitation-oriented learning

The National Institution for Academic Degrees and University Evaluation (NIAD-UE) is an exclusive institution in Japan which can award academic degrees based on the accumulation of academic credits for non-university students. An applicant who wishes to be awarded a degree from NIAD-UE must classify his obtained credits according to pre-determined criteria for each disciplinary field. The criteria are constructed by several items. A sub-committee of experts in the relevant field judges whether the course credit classification by the applicant is appropriate or not paying attention on the first item in the criteria. Recently, the Active Course Classification Support (ACCS) system has been proposed to support the sub-committee for the validation of applicants course classification. Entering a classification item number into ACCS, ACCS suggests an appropriate course that belongs to the set of classification item number. However, some difficulties of deciding appropriate item numbers still remain. This study aims to improve the method of determining item numbers, which should be judged by the sub-committee, by using machine learning. We use Exploitation-oriented Learning as the learning method for improving ACCS, and present a numerical example to show the effectiveness of our proposed method.

Textual information and correspondence analysis in curriculum analysis

Correspondence analysis has been used in text mining to deepen global understanding on the characteristics of categorized accumulated textual information, and may lead to new knowledge discovery. However, difficulty in category classification for textual information remains. This article presents a new analysis method for higher education curriculum textual information utilizing corresponding analysis with a library classification.

Web service and visualization for higher education information providing service

Important and urgent issue of higher education institutions is to provide service to their students, staffs and public with useful education information. In this paper analyzing structure of higher education information, development and improvement of the education information providing service is examined that consists of syllabus database, XML web service, curriculum analyzing system, multivariable analysis methods and other web technologies. Especially correspondence analysis is utilized in text mining in order to deepen global understanding on the characteristics of categorized and accumulated textual information included in syllabuses. This paper also presents the visualization method for higher education curriculum textual information.

Syllabus database and Web service on higher education

In this paper analyzing a structure of higher education information, we propose syllabus XML database, Web services on education information, and curriculum analyzing system based on advanced communication technology.

Mean-variance portfolio optimization model with uncertain coefficients

We deal with portfolio selection problem with interval objective function coefficients. Considering the problem as a multiple objective problem including uncertainties, two kinds of efficient solutions that are efficient or Pareto optimal for at least one or all coefficients within given intervals can be regarded as optimistic or pessimistic solutions respectively. We investigate the properties of efficiency conditions by means of two kinds of preference cones.

Possibility degree and sensitivity analysis in possibilistic multiobjective linear programming problems

We consider possibilistic multiobjective linear programming problems. Possibly efficient solution set is defined as a fuzzy set whose membership grade represents possibility degree of efficiency for a feasible solution. We restrict possibilistic distributions of coefficients in the problems, so that we can propose a novel calculation method of possible efficiency degree. Applying the result of our calculation method to sensitivity analysis, we can discuss re-adjustment for parameters of the possibilistic distribution.

An autonomous decentralized system for constraint-oriented problem solving involving continuous and fuzzy variables

A novel way of solving constraint satisfaction problems by using autonomous decentralized systems is introduced. First, the componential constraints are further decomposed into their structural and elastic parts. Second, the elastic parts are treated by interactive dynamics among autonomous decentralized problem solving processes. Third, the structural part is treated in a centralized logic based symbol processing subsystem which works concurrently and interactively with the autonomous subsystems for elastic parts. Fourth, this system architecture, is implemented on transputers by using Occam and is applied to scheduling and mechanical design problems showing the effectiveness of the approach.<<ETX>>

Structure of university database system and data analysis

This paper describes the state of research and development of data structure for university data, especially, Japanese university basic survey data and its application for business intelligence analysis. Our research and development are aimed for university data reporting and comparative analysis on the data with consideration on general university information structure. Institutional data of university, college, or college of technology are substantially important for data analysis or knowledge discovery in the higher education management field. However, university institutional data are not necessarily standardized and compiled, so it is difficult to integrate their information for various reporting and data analysis. In the past decade, a number of investigations of the integrated university database have been done to deal with various kinds of university institutional information including university survey or school basic survey data by considering the structured university data. This paper describes the state of our research and development of structure of university database and data analysis system, which are effectively utilized for data reporting and university comparative analysis. We explain (i) development of various data reporting and analysis system of university database system, and (ii) proposal of generalized layered university database structure based on Link-based data set.

Proposal of an Active Course Classification Support system with Exploitation-oriented Learning extended by positive and negative examples

The National Institution for Academic Degrees and University Evaluation is engaged in awarding academic degrees based on the accumulation of academic course credits. These course credits must be classified for designated disciplinary fields according to pre-determined criteria by an applicant. The validity of the applicants course classification is judged based on careful reading of course syllabuses by sub-committee members of the institution who are well versed in the particular field. The Active Course Classification Support system (ACCS) aims to assist subcommittee members in the course classification process. As the latest improvement of learning mechanism, Exploitation-oriented Learning (XoL) has been introduced to ACCS. The operation of ACCS has not been considered with the features of XoL, and does not consider the states of other items that are not decided by XoL. We must consider the states of all items in order to output the course properly by ACCS. In this paper, we aim to develop the effective operation of ACCS that includes the states of all items by means of generating sets of positive and negative examples derived from courses that have been judged by the subcommittee members in the past. We show a numerical example to clarify the effectiveness of our proposal.