Takumi Ichimura

Development of Coronary Heart Disease Database

We developed the Coronary Heart Disease Database for the purpose of evaluating prognostic systems. The database consists of four training data-sets and one testing dataset, each of which includes more than ten thousand records related to the development of coronary heart disease.

Density-Based Spatiotemporal Clustering Algorithm for Extracting Bursty Areas from Georeferenced Documents

Nowadays, with the increasing attention being paid to social media, a huge number of georeferenced documents, which include location information, are posted on social media sites. People transmit and collect information over the Internet through these georeferenced documents. Georeferenced documents are usually related to not only personal topics but also local topics and events. Therefore, extracting bursty areas associated with local topics and events from georeferenced documents is one of the most important challenges in different application domains. In this paper, a novel spatiotemporal clustering algorithm, called the (ϵ,τ)-density-based spatiotemporal clustering algorithm, for extracting bursty areas from georeferenced documents is proposed. The proposed clustering algorithm can recognize not only temporally-separated but also spatially-separated clusters. To evaluate our proposed clustering algorithm, geo-tagged tweets posted on the Twitter site are used. The experimental results show that the (ϵ,τ)-density-based spatiotemporal clustering algorithm can extract bursty areas as (ϵ,τ)-density-based spatiotemporal clusters associated with local topics and events.

Reasoning and learning method for fuzzy rules using neural networks with adaptive structured genetic algorithm

In this paper, we present a reasoning and learning method for fuzzy rules using neural networks with adaptive structured genetic algorithm. This adaptive structured genetic algorithm can determine the network structure and their weights solely by an evolutionary process. With this approach, no a priori assumptions about topology are needed and the only information required is the input and output characteristics of the task. The adaptive structured genetic algorithm can generate or annihilate the specified units respectively in hidden layer to achieve an overall good system, without using back propagation or any other learning algorithm.

Extraction of fuzzy rules using neural networks with structure level adaptation - verification to the diagnosis of hepatobiliary disorders

This paper presents the reasoning and learning method for fuzzy rules using structure level adaptation of neural networks. In a usual neural network mechanism, we can observe some behaviors during the learning process. Based on such behaviors of neuron activity, we can generate or annihilate the specified neurons respectively in hidden layer to achieve an overall good system. In the method that we have proposed, we have used a procedure to derive the neuron generation/annihilation automatically, and applied such a procedure to the learning system where the experimental data related to hepatobiliary disorders were used. After learning by using randomly chosen data, the proposed system correctly diagnosed over 70% of cases. According to these results, we can find that fuzzy rules have some relationship with the degree of the input weight vector. As a result, we can assume that fuzzy rules for hepatobiliary disorders are extracted from this learned network.

Discovering multiple diagnostic rules from coronary heart disease database using automatically defined groups

Much of the research on extracting rules from a large amount of data has focused on the extraction of a general rule that covers as many data as possible. In the field of health care, where people’s lives are at stake, it is necessary to diagnose appropriately without overlooking the small number of patients who show different symptoms. Thus, the exceptional rules for rare cases are also important. From such a viewpoint, multiple rules, each of which covers a part of the data, are needed for covering all data. In this paper, we describe the extraction of such multiple rules, each of which is expressed by a tree structural program. We consider a multi-agent approach to be effective for this purpose. Each agent has a rule that covers a part of the data set, and multiple rules which cover all data are extracted by multi-agent cooperation. In order to realize this approach, we propose a new method for rule extraction using Automatically Defined Groups (ADG). The ADG, which is based on Genetic Programming, is an evolutionary optimization method of multi-agent systems. By using this method, we can acquire both the number of necessary rules and the tree structural programs which represent these respective rules. We applied this method to a database used in the machine learning field and showed its effectiveness. Moreover, we applied this method to medical data and developed a diagnostic system for coronary heart diseases

Ant Colony Optimization using exploratory ants for constructing partial solutions

When Traveling Salesman Problem (TSP) is solved by Ant Colony Optimization (ACO), the round tour that each ant generated is evaluated by its tour length, and each ant lays pheromone based on the evaluated value. Basically, ants select the next city considering pheromone intensity and the closeness of the distance. In making a round tour, however, it is probable that the only far cities remain as the candidates for the next move. In this case, the ant has to select one of the cities unwillingly. By this phenomenon, a useless route is included in the round tour, and the tour is not evaluated appropriately. To solve this problem, we propose a new ACO method using heterogeneous ants for Traveling Salesman Problems. In the proposed method, there exist not only the normal ants but also the exploratory ants which construct partial solutions. In constructing solution phase, the exploratory ant selects the next city from unvisited cities which exist in the neighborhood of the ant. If there is no unvisited city in the neighborhood of the ant, the ant gives up constructing its round tour. We call this method Give-up Ant System (GAS). We confirmed that the search performance improved by the effect of the diversification of search by the exploratory ants.

Learning of neural networks with parallel hybrid GA using a royal road function

In the learning of neural networks, the hybrid genetic algorithm (GA) is one of useful methods, since it can find an optimal set of weights in shorter timer. However, the GA part requires many individuals in a population to maintain its diversity and then it remains a trade-off between the population size and time. We introduce a new idea of evaluation of its chromosome based on the building block hypothesis. We assume an index with same length of an individual and measure the length of corresponding bits to it. Then, we make a reproduction using both fitness and its new index. Furthermore, we change its length from dynamically short to long according to the convergence situation, since intermediate order schemata results from combination of the lower order schemata. To verify the effectiveness of the proposed method, we developed a medical diagnosis system. It is shown that an optimal solution was found in the population size of 10.

Fuzzy reasoning model of facial selection and its applications

We have been spending more time on interaction with computer systems in the information-oriented society. Human face-to-face interaction consists of both verbal communication and nonverbal communication such as facial expressions and gestures. Therefore we have adopted facial expressions for human computer interaction. We propose a fuzzy reasoning model for selecting faces expressing emotions caused by several situations. Nine faces with three levels of brow and eye deflection and three levels of mouth deflection were used to construct the fuzzy reasoning model. The fuzzy reasoning model was applied to selecting faces expressing emotions caused by sentences on the computer display. A questionnaire survey reveals the usefulness of our method as a human-computer interface.

A generation method of filtering rules of Twitter via smartphone based Participatory Sensing system for tourist by interactive GHSOM and C4.5

Mobile Phone based Participatory Sensing (MPPS) systems involve a community of users sending personal information and participating in autonomous sensing through their mobile phones. Sensed data can also be obtained from external sensing devices that can communicate wirelessly to the phone. We have developed the tourist subjective data collection system with Android smartphone. The tourist can tweet the information of sightseeing spots by using the application. The application can determine the filtering rules to provide the important information of sightseeing spot. The rules are automatically generated by Interactive Growing Hierarchical SOM and C4.5.

Effective Diversification of Ant-Based Search Using Colony Fission and Extinction

In ant colony optimization (ACO), to keep a balance between intensification and diversification of search is important. ASelite is one of the extensions of the original ant system (AS). In ASelite, elitist ants lay additional pheromone on the best found tour. By the intensification mechanism, ASelite can perform more rapid search for the optimal solution than conventional AS. On the other hand, however, this intensification of search causes the problem that ants are liable to fall into local optima. In this research, we aim to also improve the diversification of search while keeping the characteristic of ASelite. Therefore, we propose a new method using multiple colonies. In this method, multiple colonies search for the solution while doing colony fission and extinction. In addition, for the improvement of search performance by each single colony, we also propose the improved ASelite, ASelite with negative alpha. We applied the proposed methods to traveling salesman problems. Some experimental results show that search performance is improved and various solutions are acquired by the interaction of multiple colonies.