H. Miyagi's research while affiliated with University of the Ryukyus and other places

Society continuously communicate on social networks, blogs and news; the permanent generation of text data makes an attractive challenge to discover what people are talking about and track a non-linear status of a certain topic over the time. In this paper we explore particle filtering (PF) methodology applied to text data. PF is a powerful methodology for sequential signal processing with a wide scope of applications. There are several non-linear problems where PF has been applied, such as localization and tracking of targets, recognition of objects in video or image, among others. This paper aims to explore PF for tracking a set of related words in a news stream. We track a predefined topic and its "relevance, estimating its state value over the time. PF uses a state-space model to perform estimations, in our work; the state-space model is simulated by using Unigram method. Unigram models a topic named as "nuclear issue from a news stream composed of 421 news generated during last natural disaster in Japan. PF sampling importance re-sampling (SIR) algorithm is used to process data as it arrives, for rapid adaptation to changing signal characteristics. Our application computes the posterior pdf of the system's state based on available information; every time a news is read, the system tracks and estimates the selected topic "nuclear issue".

Migration of individuals among islands is effective in many cases in distributed parallel processing of the evolutionary computation. However, the effectiveness of migration depends on many factors such as migration frequency, migration topologies, migration scale, timing, and so on. The objective of our research is to investigate such factors for designing efficient parallel evolutionary computation. In this paper we analyzes the migration effects of the tree based island model in parallel genetic algorithms and parallel PBIL (Population Based Incremental Learning). Computational experiment shows us migration effects of the parallel evolutionary computation and influence of the migration topology and frequency.

This paper presents some results of experimen-tal evaluation on the fitness varieties in parallel genetic al-gorithms based on tree topology migrations. The evaluation focuses on the relation between the solution quality and the fitness varieties. We perform the evaluation for the typical tree topologies; the star, line, and balanced binary tree with varying migration interval since the solution quality of the parallel genetic algorithm depends on the kind of the topology and migration interval. The results show that the line topol-ogy can keep the chromosome variety of the GA and obtain better quality of solutions since the topology has the longest distance between the root and the leaves.

The purpose of this study is to derive the optimal route for sightseeing. In order to obtain the optimal route, we have defined an optimal routing problem for sightseeing (ORPS). In ORPS, the travel time or the location value has been treated as a crisp number. However, it is better that these values are expressed by ambiguous value like "about 10 minutes" rather than by crisp number. In this paper, we formalize an optimal routing problem adopting fuzzy time-varying travel time and propose an algorithm to solve the problem.

Our proposed method gives genes the priority for mutation and intersection operations. The priority of genetic locus changes by the chromosome evaluation value improvement, and shapes an effective search space area. In computational evaluation, when appropriateness priorities are given, good results are obtained. However, because the appropriate priority is not already known, we propose the dynamic priority setting method also. This method changes priority according to the improvement degree of evaluated value. This method yields fair results.

We propose a semantics filter aimed at filtering unnecessary data for a user. In the Geographic Information Systems (GIS), the transfer of spatial data for each user is one of the important problems. We design a system that filters spatial data by using user's information adoptively. Our system derives the degree of geographical knowledge by the fuzzy reasoning and decides the priority of buildings by the C4.5 using user's information. Furthermore, in the process of the priority of buildings, we determine also the priority from geometrical and topological information. This paper describes a method for determining the priority of buildings.

The purposes of the paper are to propose and evaluate an immune optimization algorithm inspired by biological immune cell-cooperation, and this algorithm solves the division-of-labor problems in a multi-agent system (MAS). The proposed algorithm solves the problem through interactions between agents, and between agents and the environment. The interactions are performed by division-and-integration processing, inspired by immune cell-cooperation and a similar co-evolutionary approach. The division-and-integration processing optimizes the work domain, and the similar co-evolutionary approach performs equal divisions. To investigate the validity, this algorithm is applied to "N-th agent's Travelling Salesmen Problem" as a typical problem of MAS. The best property for solving via MAS is clarified with some simulations

In order to solve the double-talk problem in the echo cancelling for teleconference system, the authors have proposed the extended correlation least mean squares (ECLMS) algorithm. However, this algorithm does not give a fast convergence. The purpose of this paper is to derive an orthogonal ECLMS (OECLMS) with the lattice strucure in order to speed up the convergence characteristics and to reduce the computational load for the adaptation. © 2000 IMACS. Published by Elsevier Science B.V. All rights reserved.

In this paper, a mixed sensitivity weighting strategy for H^∞ controller design for an induction generator/infinity bus windmill power system is presented. The approach is based on a form of quasi-linearization which makes the windmill power system a linear function of the control input. The system controller is then synthesized by using H^∞ standard controller design techniques. A combination of additive and multiplicative uncertainty representation is used to achieve the robust stability for a wide range of operating conditions. The capability of the proposed controller to damp out the oscillations of the power and angular velocity, and the robustness with respect to the system parameter variations and model errors are evaluated in the simulation study