We have proposed a method to partition a time-series data into several periods based on changes of trends and features, where we unify the adjacent two clusters that have the maximum total similarity. We have also proposed a method to represent a time-series data in linguistic expressions and appled to retrieval of similar time-series data using a similarity of linguistic expressions. We have linguitic expressions from some values in three fuzzy periods, called the first, second and third terms, which are fixed in the previous papers but changed based on trends and features of the data in this paper. We partion the time-series data into two periods, whose boundary is considered as the changing time from the firt to the third terms, that is, the center of the second term. We restrict it in the range of not too early and not too late times. We apply this mathod to lingusitic expressions and retrival of times-series data and we have better result than the previous ones.

National Diet, Local Assembly, Speech Analysis This paper compares the speeches made by MPs when they were part of their local assemblies and after they became members of the National Diet. The data collected from the local assembly and National Diet speeches were compared based on three aspects: remarks made about the same topics, the political vocabulary ratio (Uchida et al. 2019), and their politeness and formality with which the speeches were made. The first analysis demonstrated that MPs tend to focus on topics they previously took up in their local assemblies. The political vocabulary ratio, on the other hand, helped identify politically relevant words from those that are not. Finally, the analysis showed that MPs were relatively more polite during their National Diet speeches compared to the local assembly speeches.

Many kinds of onomatopoeia in Japanese have multiple meanings "gorogoro" is used in a different sense, such as "thunder rumbling" and "chilling out at home". From our previous research, it is known that determining the meaning of onomatopoeia is depend on the dependent verbs. However, there are also onomatopoeia without verbs. Additionally, the meaning determination may require extensive contextual information. In this paper, we examine the possibility of senses classification considering the appearance context of onomatopoeia using BERT pre-training model, which is a general-purpose language model.

We strive to develop a dialog system which have a sense of humor in order to improve user satisfaction. We have constructed a large Japanese pun database. According to the analysis of the puns, onomatopoeia words assume an important role. We analyzed funniness of Japanese puns focused on onomatopoeia words. In this article, we report the result of the analysis and discuss a Japanese pun generation method.

After-earthquake assessment of building in terms of safety and usability is performed by technicians who give their judgement based on in-field surveys and visual inspections. In this study, an efficient method is proposed for building damage detection and structural health monitoring system is introduced. A simulation was conducted using past observed earthquakes and the acceleration data of each floor is obtained. The data is divided into train set and test set. Train set is used to train a neural network which predicts each floor damage in 3 levels and test set is used for evaluating model. Also, seismic wave characteristics were investigated using principal component analysis.

In recent years, a driving investment explosion in real estate tech, that combines real estate and information technology, has been recorded. Many studies are using the benefits of real estate big data to investigate the utilization of artificial intelligence for the application of image processing and deep learning in the real estate sector. Age is an indispensable factor in determining the value of a property. Furthermore, earthquake resistance, equipment, depreciation cost, the extent of deterioration, and the building price are some of the other factors to consider when estimating the value of the property as these factors also vary with the passage of time. As the age of a building is not disclosed because of the confidentiality of property taxes, it is possible to collect this information from the registry. However, this process is time-consuming and costly when looking for multiple properties. In this study, the input images divided, we estimated the age of a building from its appearance using bagging, which is an ensemble learning method. Also, we created classification model using the convolutional neural network technique, which is a deep learning method that recognizes objects from images. The experimental result shows accuracy rate of divided images and original images have similar performance. In conclusion, proposed method of my study is not inferior compared to estimation of using deep neaural network.

According to the spread of mobile camera devices, imaging applications have been enriched. In document shooting, a highlight or a shadow used to be in the target images. Binarization method are used in most of imaging application for visibility improvement. However, it is not applicable to color documents. Contrast and gamma correction can control the overall visibility of the image, but it is hard to remove local noises. It is difficult to remove only shadows and highlights from images without altering the target information such as the color design or color tone. In this paper, we will propose the technique to achieve improved visibility without losing target information by integrating multiple captured images.

The purpose of a study was to maintain skill levels among middle traders for distinguishing the species of puffers and improve food safety and security . The models for distinguishing the species of puffers by capturing the color characteristics of dressed meat samples were developed . Meat samples from 7 species (dressed ocellate puffer , purple puffer , etc . ) were used ; these samples consisted of 110 as a basis .Discriminate models were made up of fish species as the criterion values and the color combinations as the predictor variables , respectively . An evaluation experiment using 20 specimens of each meat from 7 species correctly identified at least 93.5% of the dressed meat samples . Thus , the potential usefulness of the models was confirmed .

In this paper, we reduce the number of fuzzy rules in the deep fuzzy inference model and acquire knowledge as fuzzy rules. The number of input items used for the inference model is reduced by randomly selecting the number of input items in each layer. Therefore, it turns out that the number of rules in the whole of this model can be reduced more than that of rules in an inference model that uses all the original inputs at one time. However, in the previous model by Zhang, although the consequent part of the fuzzy rule was learned, the antecedent part was not learned. Since we need to deal with the situation where there is no prior knowledge in the problem to apply and it will be necessary to acquire knowledge from data, it is required to learn the antecedent part. In this paper, we propose a learning method for the antecedent fuzzy sets in fuzzy rules in order to obtain relationship between input and output of learning data from teach data. Then, as an example, the proposed method is applied to medical diagnosis of diabetes, the accuracy of the previous method is compared with that of the proposed method.

Recent year, scene recognition has been studied actively. Scene recognition has mainly two methods, the BoVW (bag of visual words) based method and the DNN (deep neural network) based method. However, in these methods, the component of the feature vector is represented as a local visual feature (visual words) so it is difficult to assign semantics to each component. Thus, we propose a novel scene recognition method, which is named ʻbag of object semantics.ʼOur proposed method has used the frequency of objects in a target scene as the feature vector. Each component of the vector corresponds to each object in a target scene, so our proposed method has high interpretability.

The IEEE Std. 1855-2016 defines a W3C XML-based language called Fuzzy Markup Language (FML). Researchers can define fuzzy systems independent of the software. FML is composed of knowledge base and rule base. Knowledge base describes fuzzy sets. Rule base describes fuzzy rule sets. It can be defined flexibly, and in particular, the shape of membership functions can be defined by parameters easily. Besides, a program to calculate complex membership functions is provided in FML. This paper aims to introduce FML to researchers in the field of fuzzy systems and to provide a tool to define flexible membership functions. In addition, we demonstrate the effect of the tuned membership functions on the performance of fuzzy systems.

Total Environment for Text Data Mining (TETDM) is a software for text mining, which is indexed in SOFT-CR. TETDM enables us to evaluate input data from various viewpoints through the processing tools and visualization tools. With TETDM, we analyzed recent research trends from the abstracts of the papers of some different symposiums. This analysis revealed the recent hot topics and the change of research trend associated with social change in the symposium.

We report a verification result about the knapsack problem using genetic algorithm simulator． Genetic algorithm is one of metaheuristic method that imitates evolution processes of living things. At first, we prepared baggage candidates to put it in a knapsack and investigated how evolutionary performance changed by the selection, crossover, mutation rate and difference of genetic population. In addition, we investigated the evolutionary performance when number of generations increase in expected value of selections and the relations between genetic population and calculation time.

Fuzzy reasoning is a basic method for rule-based system adopted in such fields as system control. In this research, a case study of constructing a prediction model is reported, in which a handbook and a tool software on introduction to fuzzy reasoning released in SOFT-CR web site was utilized. A rule base is constructed for revealing the connection among a residential solar power system and meteorological observations.

In this paper, we leaned a collaborative filtering using Womnavi application for running shoes. Womnavi application is a tool software for constructing recommender system utilizing questionnaire data on users’ hobbies and preferences. It is a recommendation system using collaborative filtering model by fuzzy cluster structure estimation. We developed running shoes recommendation system using Womnavi application, but the system of performance was not so good.

LogAnalyzer3 is a log analysis tool of RoboCup soccer simulation 2D league. In RoboCup, it is desirable to analyze game log files for developing a team without any human’s intuition and preconception. In this paper, we show an example of such a systematic method. Game log files include such important information for applying machine learning methods. We have developed a tool for extracting useful infor- mation in the games. LogAnalyzer3 can extract action sequences that are expressed by the gather of two continuous kick commands. In this paper, the action sequences are converted into kick distributions. This allows us to calculate dissimilarity between two action sequences.

This paper describes a real-time situation evaluation system for RoboCup Soccer Simulation 2D League and its application to a spectating system. The aim of developing the spectating system is to make the spectator’s experience of watching games more entertaining by expressing the excitement of the audience based on the game situation, and making it easy to understand the game situation. For this purpose, the evaluation value of the game situation called SituationScore is estimated in real time during the ongoing game, and the excitement of the game is expressed by the audio and visual effects according to the estimated SituationScore. How SituationScore is obtained using a deep learning method and also how a soccer monitor application is extended with audio and visual effects are explained.

In track and field, a speed gun is often used to measure the running speed of athletes. However, it is very difficult to keep measuring at the corner using that device. Therefore, in the corner, the image is checked visually by analysts. There is a problem that judgment is different by analyst and error occurs. In addition, it takes effort to check multiple videos images. In this research, we develop a system for calculating athletes' speed from video images.

In Japan, policies that deal with energetic activity of athletes and improvement in the standards of competitions are implemented towards the 2020 Tokyo Summer Olympics. From such of circumstances, studies are much conducted on applying ICT (Information and Communication Technology) to sports, in particular, on visualizing players’ positions or tracked movement using measurement instrumentation such as GNSS to analyze their performances. However, there are cases when players cannot wear measurement instrumentation during a game depending on the kind of sports. Therefore, image processing is often used as a main approach to obtaining players’ positions and tracking their movement. In this research, we propose a method for identifying and locating the players by using OpenPose, which can deal with an occlusion with only one camera by performing color interpolation on single-viewpoint image. In the experiment, we evaluate the utility of the method by verifying the identification accuracy for soccer players.

According to the guidelines on instruction by athletic clubs that the Ministry of Education, Culture, Sports, Science, and Technology established in 2013, it is important that coaches not only depend on their own experiences, but also learn from scientific studies by sports specialists. In recent years, measurement devices have been developed that can immediately give feedback on a batter’s swing; the measurement devices were further refined and were reduced in size by the development of sensor technology. The device then became able to assess the characteristics of the batter’s swing from the measurement data. However, a coaching method based on such scientific data has not yet been established in amateur baseball clubs because of the mixture of rational and practical instruction and the difficulty of the interpretation from calculated numerical values. Therefore, the purpose of this study is to examine the characteristics of a batter’s swing using an evaluation from a baseball coach with a great deal of experience when assessing batting skills.

Recently in Japan, technology for detecting the MAC address of IoT devices has progressed, and research on traffic measurement technology using Bluetooth is in progress. In particular, on expressways, implementation has been promoted, such as full-scale introduction of a time required system using Bluetooth. In this study, we examined travel time measurement technology using Bluetooth not only for automobile traffic but also for ordinary roads where various traffic phenomena such as pedestrians, bicycles, roadside facilities, etc. are mixed. . As a result of comparing with the measured traffic data observed at the same time, it has been confirmed that the travel time measurement using Bluetooth is highly applicable depending on the road condition of the ordinary road.

Convolutional Neural Networks (CNNs) achieve high discrimination performance in object recognition. Although CNNs automatically learn parameters by a data-driven method, appropriate hyperparameters must be selected to achieve high discrimination performance. However, since it is difficult to find the optimal hyperparameters by manual adjustment, studies on automatic adjustment using a search method have been actively conducted. In this research, we propose a method to optimize the number of filters in CNNs by using Genetic Algorithm (GA). Specifically, the commonly-used network architecture is used as a part of the initial population of GA for efficient search. In simulation experiments, we compare the classification accuracy of CNNs optimized by the proposed method with those by Optuna and Hyperopt.

Fuzzy Genetics-based Machine Learning (GBML) is a well-known method to design fuzzy classifiers which have high explainability. Generally, fuzzy GBML is originally designed as a batch learning. However, when focusing on IoT or data streaming, new information is generated continuously and it can be assumed that untrained classes are added after the classifier is designed based on the data at hand. In this study, we extend fuzzy GBML to learn untrained classes continuously. The proposed method enables continuous learning by discarding useless rules for the classification to generate new rules for untrained class patterns. Then, the trained patterns are reduced through a clustering algorithm in order to mitigate the increase in the computational cost. Experimental results show our method can efficiently learn untrained classes in terms of computational cost.

Recently, evolutionary multitasking that solves multiple optimization problems (tasks) in parallel using evolutionary computation has been actively studied. In evolutionary multitasking, each task has a population to be optimized by evolutionary computation. The main feature of evolutionary multitasking is that parents are selected from populations for not only its own task but also another. Our previous study showed that the search performance of evolutionary multitasking is improved by appropriately setting the frequency of crossover between different populations. However, there is no study about the effect of parent selection schemes from other tasks. Thus, selection of appropriate parents from multiple populations is not well-studied. In this paper, we focus on the similarity among individuals in the decision variable space and examine the effects of different parent selection schemes from other tasks on the search performance of evolutionary multiobjective multitasking.

In this study, we developed character coordination system using the Kansei retrieval agent model with fuzzy reasoning for acquisition user rules related to user preferences. The Kansei retrieval agent model imitates the sensitivity of each user. Previous research has demonstrated the effectiveness of the model in terms of presenting the user’s preferences by optimizing of fuzzy reasoning using numerical simulation and learning the user’s evaluation criteria. However, we have not verified the effectiveness for real users. Then, we developed a character coordination system using the model with fuzzy reasoning and extracted user’s preference rules. As a result, the developed system is effective in terms of acquiring preference rules that are the evaluation criteria of users.

In this paper, we propose estimation methods of the N-value, which is an index for measuring the ground strength. The proposed methods use multi-layered neural networks. In order to overcome the strong nonlinearities of N-value distribution for mountainous area, two types of new ideas are proposed. The one is to use an additional new feature value produced by using the geology type. The other is to use only learning data within a specific range, which is determined by using the proposed algorithm. The numerical simulation demonstrates the effectiveness of the proposed methods.

This study aimed to extract words as frequent sub-sequences from voice data given in streaming format. In the previous method, in the case of the symbol sequence which has fluctuation in the appearance interval of the symbol like voice, the network cannot be learned efficiently. In this paper, we propose an efficient learning method and aim to extract frequent subsequences faster than the previous method. In the previous method, a unit called LIF (Leaky Integrate and Fire) model is used. We proposed delaying the firing timing of this unit to adjust the connection weight more efficiently by STDP learning rule. In order to confirm the effectiveness, the length was limited to 2 symbol lengths for simplification, we measured the time which is taken to extract frequent sub-sequences by the previous method and the proposed method. By delaying the firing timing of this unit, we confirmed the spiking neural network can extract frequent sub-sequences faster than the previous method.

The differential evolution algorithm is an optimization algorithm that can easily obtain a good solution. An improvement from the viewpoint of interval estimation has been proposed, but the range of the search points and its variation in the search have not been reported in detail. In this research, we investigate about the range of the search points when applying the differential evolution algorithm to typical benchmark problems, and it is demonstrated that adjusting the parameter or changing the algorithm itself based on the range can be useful.

We have been modeled based on real soccer data analysis in RoboCup Soccer Simulation 2D. In the case of a soccer game, it is considered that soccer players share cooperative patterns in a pass scene. In this research, we analyze the positional relationship between teammate and opponent team member using Delaunay Triangulation. Furthermore, the key player which receives a pass is extracted from the gaze direction of the ball holder.

Moving object attractiveness map is a computational model that evaluates the possibility that a moving object in the field of view is gazed by the observer, which generates a view image in which each moving object is represented by shading according to the calculated gaze probability. Saliency map is, on the other hand, a computational model of visual attention that applies scientific knowledge of information processing in low-order visual cortexes to image processing, which generates an image that represents the saliency distribution based on low-level image features such as brightness and hue. This paper presents a method that combines outputs of the moving object attractiveness map and the saliency map to estimate the extent to which each area of the view image is likely to attract the visual attention of a human driver. The proposed model demonstrates that it can simulate a driver in different task engagement modes, such as watching out for surrounding moving objects or looking for a particular environmental object, by changing the weight of each of the two maps for linear combination.

In this paper an evacuation guidance system is proposed, in which an evacuation path is instructed to evacuees using a sequential sound system. In order to evaluate the performance of the evacuation guidance system the several experiments in which the subjects follow the sequential sound emitted from the array of speakers and the performance of evacuation is evaluated and discussed based on the experiments results.

With the development of Human-Machine Interaction technology, Human Computer Interaction (HCI) becomes a new research field. As a crucial part of the HCI research framework, in order to develop friendly and natural human-machine interface, EEG-based emotion recognition which aims to automatically discriminate different emotional states by using physiological signals acquired from human will be effective. Comparing to MRI, EEG has undisputed advantages both in terms of the simplicity and the cost of information collection. But in practice, it is very difficult to collect physiological data by connecting multiple electrodes because of users’ burden. In this study, in order to lighten the burden of wearing equipment, we used the EEG from less electrode to recognize the emotion of human. In the experiment, we employed DEAP dataset, a public available dataset which uses music video. Five IMF bands were extracted from the raw EEG by Empirical Mode Decomposition (EMD), and the sample entropy (SE) of each band were computed as the feature in order to train Bi-LSTM model as the binary classifier. The result showed the accuracy of 67.0% (Arousal) and 67.7% (Valence) in binary classification.

It is known that primary visual cortex uses a sparse code to efficiently represent natural scenes. Based on this fact, we built up a hypothesis that the same phenomenon happens at the higher cognitive function. Here we focus on semantic representation reflecting the meaning of words in the cerebral cortex. We applied sparse coding to the matrix consisting of paired data for both brain activity evoked by visual stimuli observed while a subject is watching a video, and distributed semantic representation made from the description of the video by means of a word2vec language model. Using this method, we obtained a dictionary matrix whose bases represent the corresponding relation between brain activity and the semantic representation. We then analyzed the characteristics of each base in the dictionary matrix. As a result, we confirmed that independent perceptual units were extracted with words representing their functional meaning.

Motor-sequence-learning (MSL) is important for rehabilitation or sports, and the effective learning procedure according to the brain activity is considered to improve MSL. In this study, we in- vestigated the effect of declarative and procedural presentations in MSL of a 3-block pattern including 3-button-sequences in hierarchical manner. After an exchanging of the order of the blocks, the alpha- power of EEG at the parietal region (C3, Cz, C4) tend to increase during MSL in procedural presentation task. In addition, in declarative presentation task, the theta-power of EEG at the forehead (F3, Fz, F4) tended to increase and intra-block-time of button press increased significantly than in procedural presen- tation task. These results suggest that types of presentation of the motor-task influenced on the brain activity and the performance of MSL.

The authors have measured electroencephalograms (EEGs) from subjects observing four images of ambiguous pictures. The equivalent current dipole source localization (ECDL) method has been applied to those event related potentials (ERPs): averaged EEGs. The paper reports the comparison results of “Saxophone player and Girl’s face.” In case of Girl’s face, reaction time on each part is earlier and process duration is shorter than that of Saxophone player. Especially in case of Girl’s face, ECDs were localized to the right and the left angular gyrus (AnG) around 370ms, and to the right post central gyrus (PstCG) around 415ms, then by way of language areas, ECDs were localized again to the right and the left AnG around 520ms. By our precedent researches, activities on the angular gyrus (AnG) are important to discriminate of the presented images.

In 2010, so-called the Arrowhead system started in the Tokyo Stock Exchange. This enables high-speed stock trading, or in the other words, high-frequency trading, in the order of milliseconds interval or less. Also, anyone can analyze the transaction data in the Tokyo Stock Exchange Data Cloud, although it is paid. In this study, we analyze these high-speed transaction data by principal component analysis. The analysis was performed in each segment, dividing the transaction time of 5 hours per day into 10 minutes. A variance-covariance matrix was calculated for each segment, the matrix was diagonalized, and the eigenvalues and eigenvectors, i.e., the principal component vectors, were calculated numerically. The features at each segments were extracted. Furthermore, we used these methods to analyze the day of the stock crash due to the rise of the Volatility Index (VIX index) and the normal calm day, and extracted the features of each. We discussed trading strategies on these days.

Price fluctuation is a random walk and the resulting statistical distribution is Gaussian. This is the starting point of financial engineering to derive, for example, the Black-Sholes Formula for derivative prices. On the other hand, it is widely known that the real price time series are not Gaussian, and the resulting statistical distribution have 'fat' tails at the both tails of the distribution. In order solve this problem, it is expected that a new knowledge and deep insight are called for. In this article, we report our new discovery in the ultra-high speed transactions in the range of a few seconds to a few minutes. For the real usage of those knowledge, it is expected to construct a new field of combined knowledge of statistics, intelligent information techniques including Fuzzy engineering.

Clustering algorithms mainly use Euclidean distance as a distance metric for calculating the similarity between patterns. However, there are various distance metrics other than Euclidean distance, and a suitable distance metric for clustering is expected to be different depending on the characteristics of data. In this paper, we perform clustering algorithms on several datasets with various metrics and examine the effect of distance metrics on their performance.

The input values of the Choquet integral models satisfy the strong commensurability. Using the property, we propose the set-function representation of the data constellation. As a set-function shows the magnitude relationship among input values, the average set function of the functions keep the property. To apply the inverse M ̈obius transformation, we can use fuzzy measure representations of the set-functions and analyze the function using fuzzy measure tools such as the Shapley values. Numerical examples are the evaluation values of an AHP model and the values of FCR method (Fuzzy-Set Concurrent Rating Method).

As multicriteria decision aid, ordinal regressions such as the conventional UTA and UTA-GMS methods were proposed. Interval UTA method has been proposed as an intermediate model between them. By applying Stochastic Multicriteria Acceptability Analysis (SMAA) which estimates probabilities of holding preferences between alternatives by generating many additive utility functions compatible to the preference information given by the decision maker, we observed that the preference relation estimated by interval utility function model identified by interval UTA is totally different from the preference proba- bilities for a certain number of alternative pairs. In this paper, modifying the evaluation function used for model estimation in interval UTA, we improve the identification method of interval utility function model so that preference evaluations by interval UTA and SMAA become similar.

Effectiveness of flowchart and PAD in expression of algorithms in programing language education has been investigated through experimental classes consisting of first year students in a faculty of education. The rate of students understanding and coding correctly algorithms expressed by flowchart is 35 % and that in the case of PAD is 100 %, although only five students have chosen the algorithms expressed by PAD. This possibly suggests that PAD is superior to flowchart in effectiveness of learning support for beginners of programing.

When humans get ambiguous information (numerically or linguistically), there may not be a correct answer to the question of how to process it. In the first step of this lecture, some studies conducted by psychologists are introduced. Next, the models using fuzzy logic which I worked on so far, are introduced and each of the outline are explained. In the models of human behavior, no matter how theoretically the model is refined, it does not necessarily match the experimental results. On the other hand, the model and the experimental results may fit more than expected. Here, I would like to introduce a free-flowing research process, such as the repetition of various mistakes and the appearance of another model from the idea of one model.

Spread of electrical social data leads to prevailing public open databases. This article considers the application of canonical correlation analysis to open data of financial data of educational institutions, and discusses the some issues of the data analysis.

Rough set theory was proposed by Z.Pawlak in 1982. This theory can mine knowledge through a decision rule from a database, a web base, a set and so on. The decision rule is used for data analysis as well. And we can reason for an unknown object using the decision rule. However, there are cases where too many decision rules are found. It is difficult to acquire knowledge from too many rules. However, the C.I. value of the merged decision rule may be low. In this paper, we propose a method to reduce the number of rules by merging decision rules. It is difficult to acquire knowledge from many rules. It is also difficult to find knowledge from rules with a large number of condition attributes. We propose a method to reduce condition attributes. Change the number of condition attribute categories and verify whether the number of merge rules decreases.

VRIO analysis is the method used to analyze firm’s internal resources and capabilities to find out if they can be a source of competitive advantage: the question of Value, the question of Rarity, the question of Imitability, and the question of Organization. The basic strategic process that any firm goes through begins with a vision under the philosophy, and conduct internal and external environment analysis, the most effective strategic choices, and strategic implementation. In this time, VRIO analysis is used the internal environment analysis. This research apply the fuzzy theory to the VRIO analysis, and discuss the useful of Fuzzy VRIO analysis.

China’s medical device market has significant growth and huge market size. In recent years, the demand for medical device and technology has gradually increased. If Japanese medical device companies enter the China’s market they can be greatly developed. This study deals with the advance of Olympus endoscopes in China. And, the competitive advantage of Olympus in China where stomach diseases occur frequently is clarified using fuzzy VRIO analysis.

An interval-type regression model illustrates the possibilities of an analyzed system according to its intervals that include samples. However, when the vagueness of an analyzed system is large, an obtained regression is distorted. In this case, an analyzed system may be understood from areas where sample density is high. Therefore, the author has built an interval-type regression that illustrates the principal characters of an analyzed system. In addition, the model was confirmed its usefulness using a numerical example. This paper discusses about the proposed model.

Clustering is the process of organizing similar objects into groups, with its main objective of organizing a collection of data items into some meaningful groups. The problem of clustering has been approached from different disciplines during the last few years. Many algorithms have been developed in recent years for solving problems of numerical and combinatorial optimization problems. Most promising among them are swarm intelligence algorithms. Clustering with PSO is emerging as an alternative to more conventional clustering techniques because of its capability to find global optima. Fuzzy clustering with PSO algorithm have recently been shown to produce good results in a wide variety of real-world data. In this paper, a survey on fuzzy clustering based on optimization of objective function with PSO is described.

In urban areas that have been severely damaged by disasters, it is important to set up a temporary road route to remove the rubble of completely destroyed houses. In such a case, in order to remove them as efficiently as possible, the shape of the route should be a straight line, and the route should be set so that the total of the distances from each house to the temporary road is minimized. Furthermore, it is better to be able to change the number of routes depending on the cost of installing a temporary road. Setting up convenient temporary roads for all houses is important, however, considering the houses in isolated places (isolated houses), the convenience to other houses may be significantly impaired. In this paper, we focus on the clustering method by objective function optimization, especially Fuzzy c-varieties (FCV), and work on the route setting of the temporary road by using FCV. FCV is a method to classify data into a given number of linear varieties, and has high affinity to this application. Furthermore, in order to decide the treatment of isolated houses, we propose a clustering algorithm that introduces noise terms. We also examine each method through numerical examples.