Search

In this paper, consecutive meals planning is formulated as a multiset iteration permutation problem that determines the optimal meals plan on a period consisting of consecutive days. In this problem, a meal is characterized by some characteristics such as food style, ingredient, cooking method, and so on. The evaluation function is defined by use of information entropy for measuring the appearance order of meal's characteristics on the meals plan. For optimizing this problem, Genetic Algorithms (GAs) with escape from stagnation of search are proposed. It is empirically shown that the proposed GAs with escape operation work better than the conventional permutation GA without escape operation for small and large size optimization problems. [ABSTRACT FROM AUTHOR]

SUMMARY In recent years, a study of evolvable hardware (EHW) which can adapt to new an unknown environment attracts much attention among hardware designers. EHW is reconfigurable hardware and can be implemented combining reconfigurable devices such as FPGA (Field Programmable Gate Array) and evolutionary computation such as Genetic Algorithms (GAs). As such research of EHW, Block-Based Neural Networks (BBNNs) have been proposed. BBNNs have simplified network structures and their weights and network structure can be optimized at the same time using GAs. The learning of BBNNs without constraint of network structure is, however, not efficient because the degree of difficulty of learning depends on network structures. In this paper, we proposed a new evaluation index of network structures for BBNNs based on the least number of routes which are from inputs to outputs, and apply it to the structure search. The learning of BBNNs is efficiently executed with structure constraint condition based on the proposed index because the network structures which are difficult to learn are excluded. In order to evaluate the proposed method, we apply it to XOR, 3 bit-parity, square function approximation, contact lenses fitting, Fisher's iris classification, and Wine classification. Results of computational experiments indicate the validity of the proposed method. [ABSTRACT FROM AUTHOR]

This paper proposes a new diploid operation technique using probability for function optimization in nonstationary environments and describes a feature of diploid genetic algorithms (GAs). The advantage of the technique over previous diploid GAs is that one genotype is transformed into many phenotypes based on probability. This transformation is not made at random. It has a certain range of probabilities. Each individual has a range. The range allows adaptation to various environments. The technique allows genes to give a probabilistic representation of dominance, and can maintain the diversity of individuals. The experimental results show that the technique can adapt to severe environmental changes where previous diploid GAs cannot adapt. This paper shows that the technique can find optimum solutions with high probability and that the distribution of individuals changes when the environment changes. In addition, by comparing the proposed diploid GA with a haploid GA whose chromosome is twice the length, the features of the diploid are described. © 2010 Wiley Periodicals, Inc. Electron Comm Jpn, 93(8): 38–46, 2010; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10097 [ABSTRACT FROM AUTHOR]

SUMMARY: In this paper, we try to acquire various behavior patterns of autonomous exploration agent using several learning environments. In case of previous learning methods using a single behavior rule set, it is hard to acquire the behavior that covers all learning environments. In our method, we divide learning environments into some primitive environments whose properties differ each other, and then generate modules that are specialized for each primitive environment. To optimize behavior rules of agents, we adopt graph structured program evolution (GRAPE) which can automatically generate graph structured programs. In unknown environments, each module is switched by a program named “switcher”. The switcher selects the module that acts better in a neighboring environment. Through several experiments, our method achieved higher exploration rate in unknown environments compared to simple GRAPE, random search, and the method that switches modules randomly. [ABSTRACT FROM AUTHOR]

The construction of image transformation manually requires large amounts of effort, and thus several methods of automating it with machine learning, such as neural networks or genetic programming, have been proposed. Most of these methods are just constructed image filters that calculate an output value from values in the local area of each pixel independently. However, in several tasks, such as area detection, information on more distant areas is helpful to processing. In this paper, we introduce a new neural network model for automatic construction of image transformations. The proposed model is composed of a regular array of identical evolutionary neural networks, the Real-Valued Flexibly Connected Neural Networks (RFCN) that we previously proposed, and each RFCN is connected to neighboring RFCNs. The proposed model is called Cellular RFCN (CRFCN). Because of the local connections, each RFCN can consider information on distant areas indirectly. We apply CRFCN to three image transformation tasks, compare it with other methods, and examine its effectiveness. © 2013 Wiley Periodicals, Inc. Electron Comm Jpn, 96(5): 17-27, 2013; Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/ecj.11460 [ABSTRACT FROM AUTHOR]

The purpose of this paper is to optimize the structure of hierarchical neural networks. In this paper, structure optimization is used to represent a neural network by the minimum number of nodes and connections, and is performed by eliminating unnecessary connections from a trained neural network by means of a genetic algorithm. We focus on a neural network specialized for image recognition problems. The flow of the proposed method is as follows. First, the Walsh-Hadamard transform is applied to images for feature extraction. Second, the neural network is trained with the extracted features based on a back-propagation algorithm. After neural network training, unnecessary connections are eliminated from the trained neural network by means of a genetic algorithm. Finally, the neural network is retrained to recover from the degradation caused by connection elimination. In order to validate the usefulness of the proposed method, face recognition and texture classification examples are used. The experimental results indicate that a compact neural network was generated, maintaining the generalization performance by the proposed method. © 2012 Wiley Periodicals, Inc. Electron Comm Jpn, 95(3): 28-36, 2012; Published online in Wiley Online Library (). DOI 10.1002/ecj.10384 [ABSTRACT FROM AUTHOR]

Keyword image retrieval is now widely studied. By using such technologies, we can obtain images with the corresponding keywords easily. In the case of conventional image search systems, we basically search according to file names. However, the file names given are frequently incorrect. To resolve this problem, we propose an automatic keyword addition method for scenery images. In this paper, there are two important points. One is the image segmentation method using the maximum distance algorithm (MDA). The other is automatic keyword addition using the color features of regions. In the image segmentation method, we propose an automatic decision method for the parameters of the MDA. For this purpose, we investigate the relation between the optimal parameters and the features of regions. For the color feature extraction of regions, we propose a genetic algorithm (GA). Moreover, in order to show the effectiveness of the proposed method, we provide simulation examples. The results of simulations demonstrate the effectiveness of keyword addition for scenery images. © 2012 Wiley Periodicals, Inc. Electron Comm Jpn, 95(2): 39-49, 2012; Published online in Wiley Online Library (). DOI 10.1002/ecj.10364 [ABSTRACT FROM AUTHOR]

This paper introduces a novel statistical model for estimating the intraindividual difference in left prefrontal cortex electroencephalogram (EEG) activity, and a method for evaluating the proposed model. It is known that EEGs contain individual characteristics. However, extraction of these individual characteristics has not been reported. The analyzed frequency components of an EEG can be subdivided into components that contain a significant number of features, and components that do not contain such features. From the viewpoint of these feature differences, we propose a model for extracting the features of an EEG. The model assumes a latent structure and employs factor analysis, treating the model error as personal error. We consider the first factor loading, which is calculated by eigenvalue decomposition, as the EEG feature. Furthermore, we use a k-nearest neighbor (kNN) algorithm for evaluating the proposed model and the extracted EEG features. In general, the distance metric used is the Euclidean distance. It is possible that the distance metric used depends on the characteristics of the extracted EEG features and on the subject. Therefore, depending on the subject, we use one of three distance metrics: the Euclidean distance, the cosine distance, or the coefficient of correlation. Finally, in order to show the effectiveness of the proposed model, we present the results of an experiment using real EEG data. © 2011 Wiley Periodicals, Inc. Electron Comm Jpn, 94(5): 9-16, 2011; Published online in Wiley Online Library (). DOI 10.1002/ecj.10326 [ABSTRACT FROM AUTHOR]

This paper presents a new self-position estimation method for an autonomous mobile robot whose processing time can be varied. In this method, the current position of the robot is estimated by image template matching with the normalized correlation coefficient between the input image and stored images. Based on the idea of anytime sensing, the time for self-position estimation can be varied by changing the image size. In order to realize efficient self-position estimation, image templates are generated by means of Genetic Algorithm. The usefulness of the proposed method is shown through simulation results using test images and experimental results obtained with a real robot. © 2010 Wiley Periodicals, Inc. Electron Comm Jpn, 93(11): 46-58, 2010; Published online in Wiley Online Library (). DOI 10.1002/ecj.10223 [ABSTRACT FROM AUTHOR]

It is well known that index funds are popular passively managed portfolios and have been used very extensively in hedge trading. Index funds consist of a certain number of stocks of listed companies on a stock market such that the fund's return rates follow a similar path to the changing rates of the market indices. Thus, index fund optimization can be viewed as a combinatorial optimization problem for portfolio management. In this paper, we propose an optimization method that consists of a genetic algorithm and a heuristic local search algorithm to make strong linear association between the fund's return rates and the changing rates of the market index. We apply our method to the Tokyo Stock Exchange and create index funds whose return rates follow a similar path to the changing rates of the Tokyo Stock Price Index (TOPIX). The results show that our proposed method creates index funds with a strong linear association to the market index with minimal computing time. © 2010 Wiley Periodicals, Inc. Electron Comm Jpn, 93(10): 42-52, 2010; Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/ecj.10099 [ABSTRACT FROM AUTHOR]

In this paper, we propose a hybrid genetic algorithm with fuzzy logic controller (flc-rkGA) to solve the resource-constrained multiple project scheduling problem (rc-mPSP) which is well known as an NP-hard problem and the objective in this paper is to minimize total complete time in the project. It is difficult to treat the rc-mPSP problems with traditional optimization techniques, The new approach proposed is based on the hybrid genetic algorithm (flc-rkGA) with fuzzy logic controller (FLC) and random-key encoding. For these rc-mPSP problems, we demonstrate that the proposed flc-rkGA to solve the rc-mPSP problem yields better results than several heuristic genetic algorithms presented in the computation result. [ABSTRACT FROM AUTHOR]

SUMMARY The method that binarizes an image by local thresholds in separated image areas is useful when the image has uneven brightness. A method was proposed to separate an image dynamically for local thresholding to generate a binary image by using a genetic algorithm (GA). However, the existing method used only horizontal and vertical lines for image separation. This paper suggests a method of separating an image dynamically using diagonal lines for local thresholding in order to generate a binary image by using a GA and of evaluating binary images. The experimental results show that the images binarized by the proposed method have good separation of objects from the background, and that they include less noise and blur than those binarized by the existing method. [ABSTRACT FROM AUTHOR]

SUMMARY Differential evolution (DE) is an effective optimization method for global continuous optimization problems. Recently, we developed local descent direction vector based differential evolution (LDDVDE), which uses local descent direction vectors based on the operation vectors in order to improve the local search performance of DE. In this paper, we extend LDDVDE to multiobjective optimization problems. We adopt the hyper-volume indicator to order the operation vectors to make the local descent direction vectors for the case of multiobjective optimization problems. The effectiveness of the proposed method is confirmed through some numerical experiments for typical benchmark problems. [ABSTRACT FROM AUTHOR]

Standing from a chair is an important task of daily living for physically handicapped people. In a rehabilitation center, a healthcare professional plans motion on the basis of experience and knowledge so that a patient may stand up with little load. Therefore, there is a problem that the plan is occasionally different with each healthcare professional. In this paper, a method for generation of a motion trajectory to stand from a seated position with little load by using a genetic algorithm (GA) is proposed. The human body is expressed as a three-rigid-link model. In the model, the ankle, the knee, and the waist are set as the joints. Electromyographic (EMG) measurements of the muscle driving each joint were made and a model relating each joint torque to the EMG was constructed using the ARX model. The motion trajectory to stand from a seated position was generated by using a GA with its evaluation function based on the constructed ARX model. The generated motion trajectory was evaluated by experimental work with eight healthy subjects. The effect of the proposed method was objectively verified by the subjects' EMGs. In addition, the subjective effect of the proposed method was verified by analysis of the variance of the subjects' impressions. © 2012 Wiley Periodicals, Inc. Electron Comm Jpn, 95(9): 36-43, 2012; Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/ecj.11395 [ABSTRACT FROM AUTHOR]

Interconnected recurrent neural networks are well-known, with their abilities of associative memory of characteristic patterns. For example, the traditional Hopfield network (HN) can recall stored patterns stably, and Aihara's chaotic neural network (CNN) is able to realize dynamical recollection of a sequence of patterns. In this paper, we propose to use meta-heuristic (MH) methods such as particle swarm optimization (PSO) and the genetic algorithm (GA) to improve traditional associative memory systems. Using PSO or GA, for CNN, the optimal parameters are found to accelerate the recollection process and raise the rate of successful recollection, and for HN, the optimized bias current is calculated to improve the network with dynamical association of a series of patterns. Simulations of binary pattern association showed the effectiveness of the proposed methods. © 2012 Wiley Periodicals, Inc. Electron Comm Jpn, 95(6): 12-23, 2012; Published online in Wiley Online Library (). DOI 10.1002/ecj.11372 [ABSTRACT FROM AUTHOR]

This paper deals with formulating a reusable reverse logistics model considering decisions on backorder/next arrival of goods. The optimization method for minimizing the transportation cost and minimizing the volume of backorder or next arrival of goods by Just In Time (JIT) delivery at the final delivery stage between the manufacturer and the processing center is proposed. The optimization algorithm proposed combines priority-based encoding/decoding and the hybrid genetic algorithm with a fuzzy logic controller, and suboptimal delivery routes are determined. Based on a case study of a distilling and sales company in Busan in Korea, solution by the new model of the reusable reverse logistics of empty bottles is performed and the effectiveness of the proposed method is verified. © 2012 Wiley Periodicals, Inc. Electron Comm Jpn, 95(5): 42-55, 2012; Published online in Wiley Online Library (). DOI 10.1002/ecj.10387 [ABSTRACT FROM AUTHOR]

In this paper, we propose a road traffic control method for reducing traffic congestion with a genetic algorithm. In the not too distant future, a system which controls the routes of all vehicles in a certain area must be realized. The system should optimize the routes of all vehicles, but the solution space of this problem is enormous. Therefore, we apply the genetic algorithm to this problem by encoding the route of all vehicles to a fixed length chromosome. To improve the search performance, a new genetic operator called 'path shortening' is also designed. The effectiveness of the proposed method is shown experimentally. © 2012 Wiley Periodicals, Inc. Electron Comm Jpn, 95(4): 11-19, 2012; Published online in Wiley Online Library (). DOI 10.1002/ecj.10421 [ABSTRACT FROM AUTHOR]

Particle swarm optimization (PSO) is one of the most powerful methods for solving unconstrained and constrained global optimization problems. A cluster-structured PSO with interaction and adaptation is proposed in this paper, and the cluster structure, interaction, and adaptation of the proposed PSO are analyzed by numerical simulations. The feasibility and advantages of the proposed cluster-structured PSO are demonstrated by numerical simulations using some typical global optimization test problems. © 2011 Wiley Periodicals, Inc. Electron Comm Jpn, 94(11): 9-17, 2011; Published online in Wiley Online Library (). DOI 10.1002/ecj.10379 [ABSTRACT FROM AUTHOR]

In normalized correlation matching, a template image is set by manual operation before the matching process. Namely, the contents and the size of a template image are determined by the human sense. This paper proposes a method of performing high-speed normalized correlation matching by extracting multiple partial areas automatically, which is effective in image matching. These extracted multiple partial areas become the new template image. The proposed method extracts multiple partial areas suitable for matching by genetic algorithm. The experimental results show that multiple partial areas including an image pattern that was useful for matching were extracted by the proposed method and that the processing time for image matching was reduced to 50%. The proposed method has higher reliability than conventional methods. © 2011 Wiley Periodicals, Inc. Electron Comm Jpn, 94(10): 1-9, 2011; Published online in Wiley Online Library (). DOI 10.1002/ecj.10373 [ABSTRACT FROM AUTHOR]

The Uncapacitated Facility Location Problem (UFLP) is a fundamental optimization problem involving the selection of locations at which facilities supplying the same service are to be placed. Since it has been shown that the UFLP is NP-hard, it has generally been thought that there is no hope of finding a polynomial time algorithm by which an optimal solution is always obtained. In this paper, we propose a genetic algorithm for solving the UFLP. In the UFLP, according to the ratio of the cost of facility placement and the cost to users of the facility, the number of facility locations can be roughly estimated. Therefore, partial solution spaces that are likely to contain a good solution can be predicted to some extent on the basis of the classification index. By using mutation with the operation that searches the solution space that is likely to contain a good solution, the proposed method can search the whole space of solutions efficiently. Its effectiveness is shown by a numerical experiment in which our method is compared with existing methods. © 2011 Wiley Periodicals, Inc. Electron Comm Jpn, 94(5): 47-54, 2011; Published online in Wiley Online Library (). DOI 10.1002/ecj.10180 [ABSTRACT FROM AUTHOR]

In this paper, an improved guided genetic algorithm is proposed for the job-shop scheduling problem. The proposed method is improved by a genetic algorithm using multipliers which can be adjusted during the search process. Simulation results based on some benchmark problems demonstrate that the proposed method can find better solutions than the genetic algorithm and the original guided genetic algorithm. © 2010 Wiley Periodicals, Inc. Electron Comm Jpn, 93(8): 15–15, 2010; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10263 [ABSTRACT FROM AUTHOR]

In order to efficiently obtain an approximate solution of the traveling salesman problem (TSP), the use of extended changing crossover operators (ECXO) which can substitute for any crossover operator of genetic algorithms (GAs) and ant colony optimization (ACO) for another crossover operator at any time is proposed. In this investigation, ECXO uses both ACO or edge recombination crossover (EX) and edge exchange crossover (EXX) in the early generations in order to create local optimum subpaths, and it uses edge assembly crossover (EAX) to create a global optimum solution after a certain number of generations. In EX or ACO any individual or any ant determines the next city it visits from the lengths of the edges or the amounts of pheromone deposited on the edges, where the pheromone indicates the lengths of tours that other ants have completed, thus generating local optimum paths. In EXX, the generated paths converge to a provisional optimal path. In EAX, a parent exchanges edges with another parent to create subcyclic paths, before restructuring the total cyclic path by combining the subcyclic paths so as to minimize the distances between them. In this paper the validity of ECXO is experimentally verified by using medium-sized problems from TSPLIB, and it is shown that ECXO can find the best solution earlier than EAX. © 2010 Wiley Periodicals, Inc. Electron Comm Jpn, 93(7): 1–16, 2010; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10313 [ABSTRACT FROM AUTHOR]

The logistics network design is one of the important phases of Supply Chain Management (SCM) and it is a problem that should be optimized for long-term promotion of efficiency of the whole supply chain. Usually a plant produces different types of products. Even if it is a factory of the same company, delivery differs according to the kind of produced product. The restrictions which this model has are deeply concerned with TP in the real world. In this paper, we consider the logistics network design problems with multiple products and constraints for delivery course. To solve the problem, we used a hybrid priority-based genetic algorithm (h-priGA), and we tried comparison experiments with priority-based genetic algorithm (priGA) and h-priGA, and show the effectiveness of h-priGA. [ABSTRACT FROM AUTHOR]

This paper proposes a method to distinguish the key-type of human fingers attached to small permanent magnets. The Hall sensors arrayed in the credit card-size area feel the distribution of the magnetic field due to the key-typing movement of the human fingers as if a keyboard exists, and the signal is analyzed using the genetic algorithm or the neural network algorithm to distinguish the typed keys. By this method, the keyboard can be miniaturized to credit card size (54 mm × 85 mm). We called this system “the virtual keyboard system.” © 2009 Wiley Periodicals, Inc. Electron Comm Jpn, 92(3): 32–37, 2009; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10043 [ABSTRACT FROM AUTHOR]

EEG is characterized by unique and individual characteristics. Little research has been done to take into account the individual characteristics when analyzing EEG signals. Often the EEG has frequency components which can describe most of the significant characteristics. Then there is the difference of importance between the analyzed frequency components of the EEG. We think that the importance difference shows the individual characteristics. In this paper, we propose a new EEG extraction method of characteristic vector by a latency structure model in individual characteristics (LSMIC). The LSMIC is the latency structure model, which has personal error as the individual characteristics, based on normal distribution. The real-coded genetic algorithms (RGA) are used for specifying the personal error that is unknown parameter. Moreover we propose an objective estimation method that plots the EEG characteristic vector on a visualization space. Finally, the performance of the proposed method is evaluated using a realistic simulation and applied to real EEG data. The result of our experiment shows the effectiveness of the proposed method. © 2009 Wiley Periodicals, Inc. Electron Comm Jpn, 92(1): 9–17, 2009; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10009 [ABSTRACT FROM AUTHOR]

Researchers on action control of autonomous agents and multiple agents have attracted increasing attention in recent years. The general methods using action control of agents are neural network, genetic programming, and reinforcement learning. In this study, we use neural network for action control of autonomous agents. Our method determines the structure and parameter of neural network in evolution. We proposed Flexibly Connected Neural Network (FCN) previously as a method of constructing arbitrary neural networks with optimized structures and parameters to solve unknown problems. FCN was applied to action control of an autonomous agent and showed experimentally that it is effective for perceptual aliasing problems. All of the experiments of FCN, however, are only in grid space. In this paper, we propose a new method based on FCN which can decide correction action in real and continuous valued space. The proposed method, called Real-valued FCN (RFCN), optimizes input–output functions of each unit, parameters of the input–output functions and speed of each unit. In order to examine its effectiveness, we applied the proposed method to action control of an autonomous agent to solve continuous-valued maze problems. © 2008 Wiley Periodicals, Inc. Electron Comm Jpn, 91(2): 31–39, 2008; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.10032 [ABSTRACT FROM AUTHOR]

This paper proposes a global optimization method based on thoroughly searching for equilibrium points of gradient-based dynamical systems. Such a method is possible due to the linkage between equilibrium points of nonlinear systems and the outstanding properties of gradient dynamics. As the essence of this study, a general form of computational procedure for efficiently finding equilibrium points of nonlinear dynamical system based on the use of trajectories initiating from already known points into their eigendirections is provided. Then, optimization is realized by incorporating the procedure to gradient-based models for obtaining various local optima as their stable equilibrium points. Its application to constrained global optimization is also discussed, and the effectiveness of our method is demonstrated through numerical simulations. © 2008 Wiley Periodicals, Inc. Electron Comm Jpn, 91(1): 19– 31, 2008; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.10026 Copyright © 2008 [ABSTRACT FROM AUTHOR]

SUMMARY This paper proposes novel artificial bee colony (ABC) algorithms for solving problems including interdependence among variables. ABC algorithms are one method of solving multivariable real number space optimization problems, in which the search space is a set of vectors constructed of variables. The main search process in the ordinary ABC algorithm creates a new solution vector by changing only one variable of the current solution vector. Therefore, the new solution vector is created along only one coordinate axis. This procedure, however, is not appropriate for solving problems including interdependence among variables. For such problems, a method that is able to change more than one variable of a solution vector at the same time is required. In our proposed methods, the original coordinate axes are transformed to linearly uncorrelated axes by using principal component analysis (PCA) in the searching process. Our ABC algorithms create a new solution vector along one of the axes transformed by PCA. Hence, from the viewpoint of the original coordinate axes, the new algorithms are able to change more than one variable. The proposed algorithms have been compared with the ordinary ABC algorithm by solving five benchmark problems. Through the computer simulation results, our algorithms were shown to have better performance for solving problems including interdependence among variables than the ordinary ABC algorithm. [ABSTRACT FROM AUTHOR]

SUMMARY A complex network design method that finds a desired network structure can be a powerful tool in large-scale system design. Conventional complex network design methods tackle only static networks, that is, they do not consider growth of the target network. In this paper, we propose a new method for the design of a growing complex network. First, we consider evaluation functions which quantitatively represent the characteristics of desired structures by using feature quantities. Then, we formulate the design problem of a growing complex network as a multi-objective optimization problem in order to determine the connection targets of new nodes by using the evaluation functions. By solving the problem, we grow the network, thus obtaining the desired network. We try to generate networks which have the desired clustering coefficient and average path length concurrently. Through numerical experiments, we confirm that the proposed method is effective as a method for the design of growing complex networks. © 2013 Wiley Periodicals, Inc. Electron Comm Jpn, 97(1): 70-81, 2014; Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/ecj.10382 [ABSTRACT FROM AUTHOR]

The logistics model used in this study is a three-stage model employed by an automobile company, which aims to solve traffic problems at a total minimum cost. Recently, research on the metaheuristics method has advanced as an approximate means for solving optimization problems as found in this model. These problems can be solved using various methods such as a genetic algorithm (GA), simulated annealing, and tabu search. A genetic algorithm is superior in terms of robustness and adjustability toward a change in the structure of these problems. However, a genetic algorithm has a disadvantage in that it has a slightly inefficient search performance because it carries out a multipoint search. A hybrid genetic algorithm that combines another method is attracting considerable attention since it can compensate for a fault in a partial solution in which early convergence has a negative impact on a result. In this study, we propose a novel hybrid random key-based genetic algorithm (h-rkGA) that combines local search and parameter tuning of the crossover rate and mutation rate; h-rkGA is an improved version of the random key-based genetic algorithm (rk-GA). We attempted comparative experiments with a spanning tree-based genetic algorithm, priority-based genetic algorithm, and random key-based genetic algorithm. Further, we attempted comparative experiments with 'h-GA by only local search' and 'h-GA by only parameter tuning.' We report the effectiveness of the proposed method on the basis of the results of these experiments. © 2012 Wiley Periodicals, Inc. Electron Comm Jpn, 95(5): 56-65, 2012; Published online in Wiley Online Library (). DOI 10.1002/ecj.10374 [ABSTRACT FROM AUTHOR]

SUMMARY The development of downsized digital cameras has caused numerous undesirable shaking-blurred images. In order to restore these images, we need to estimate the PSF (point spread function) from them. It has been proposed to estimate the PSF using cepstrum images. There are PSF features in the cepstrum images of blurred images. We can estimate the PSF by searching pixels in the cepstrum images. However cepstrum images also show features of ground-truth images, so that we cannot estimate the PSF accurately. We propose a method of PSF estimation using a genetic algorithm (GA). We use the cepstrum images as a fitness function, so that the estimation results are not affected by other features in the cepstrum images. The effectiveness of our proposed method is confirmed by simulations. [ABSTRACT FROM AUTHOR]

An artificial ecosystem model was developed to study complex behavior of real organisms. The subject of the research is multigenerational migration of the Monarch butterfly in North America, and we approached the problem from an evolutionary simulation. The ecosystem consists of artificial agents and five areas. The model is based on real organisms and the areas where they live simulated with a genetic algorithm. We then designed a model of agents that have genetic factors. These genetic factors determine the behavioral strategies, physical features, and transformational strategies of the agent. Each area is modeled on areas where we can see the migration of the Monarch, and the environmental factors of each area change periodically. Because of the genetic factor of the agent and the change of the environmental factors, the agent adapts to the environment and evolves gradually by using a genetic algorithm. Results of the evolutionary simulation show that multigenerational migration behavior of the agent emerges as its genetic factors adapt to periodic changes of the environmental factors in their evolution. The migration process of agents and their genetic factors are discussed, and the migration of the agent and that of the Monarch butterfly are compared. [ABSTRACT FROM AUTHOR]

Recently, a system using bug-type artificial life was proposed for function discovery, and has been improved further. This system is an extended model of GA and GP. However, when the observation data is very complicated, the function is occasionally not obtained. A new concept is now introduced so that the function search can be applied to complicated observation data. The function search by the S-System is performed two or more times. This is termed search accumulation. To confirm the validity of search accumulation, the Himmelblau function, the valley function, and equal loudness level contours (ISO 226) are used as observation data. Since the distributions of the data are complicated, it is difficult to express them as an approximation function. By the use of the search accumulation strategy, a function that is in good agreement with the distribution can be successfully obtained. Thus, the validity of this strategy is confirmed. Search accumulation is also applicable to GP. © 2010 Wiley Periodicals, Inc. Electron Comm Jpn, 93(10): 53-62, 2010; Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/ecj.10098 3 [ABSTRACT FROM AUTHOR]

Recently, genetic network programming (GNP), an extension of the genetic algorithm (GA) and genetic programming (GP), has been proposed. GNP can make compact programs and can memorize the past history implicitly, because it expresses the solution by directed graphs and therefore can reuse the nodes. In this research, intron-like nodes are introduced for improving the performance of GNP. The aim of introducing intron-like nodes is to use every node as much as possible. It is found from simulations that intron-like nodes are useful for improving the training speed and generalization ability. © 2010 Wiley Periodicals, Inc. Electron Comm Jpn, 93(8): 23–31, 2010; Published online in Wiley InterScience (www.interscience. wiley.com). DOI 10.1002/ecj.10259 [ABSTRACT FROM AUTHOR]

We discuss a solution method based on evolutionary technology for the optimal component allocation problem in a series-parallel redundant system. A series-parallel system consists of subsystems that are connected in series and each subsystem consists of interchangeable components in parallel. There are some heuristic methods to approximately solve the optimal component allocation problem for series-parallel systems. We have formulated this problem as a multi-objective optimization problem minimizing the system cost and maximizing the system reliability, and proposed an algorithm that obtains the exact solutions (Pareto solutions) of the problems in an efficient way. Because this problem is one of the NP-complete problems, it is difficult to obtain the optimal solution for the large-scale problems and methods that obtain the exact solutions are not known. The algorithm utilizes the depth-first search method to eliminate useless searches and employs the branch-and-bound method to obtain the Pareto solutions. According to the results of our numerical experiments, the algorithm searches the Pareto solutions in practical execution time for not-so-large-scale problems. In order to solve larger-scale problems, we propose a multi-objective genetic algorithm (MOGA). We evaluate the ability of the MOGA by comparison with the exact solution method by using various scale problems. Through those experiments, we discuss the characteristics of this problem and analyze the effectiveness of our method. © 2009 Wiley Periodicals, Inc. Electron Comm Jpn, 92(9): 7–16, 2009; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10100 [ABSTRACT FROM AUTHOR]

Information on travel time is useful in improving traffic efficiency, such as reducing traffic congestion effectively. Travel time is computed from the passage time of vehicles observed at both the origin (upstream) and destination (downstream) points. For this purpose, we are .studying a vehicle identification method using vehicle height data obtained from ultrasonic vehicle detectors. However, matching methods using only the vehicle height suffer from poor identification accuracy because many similar vehicles exist on the road. Therefore, we propose a method based on matching vehicle .sequences (two or more vehicles). In addition, we develop this method using a Genetic Algorithm based on traffic characteristics. [ABSTRACT FROM AUTHOR]

To build a cooperative logistics network covering multiple enterprises, a planning method that can build a long-distance transportation network is required. Many strict constraints are imposed on this type of problem. To solve these strict-constraint problems, a selfish constraint satisfaction genetic algorithm (GA) is proposed. In this GA, each gene of an individual satisfies only its constraint selfishly, disregarding the constraints of other genes in the same individuals. Moreover, a constraint prechecking method is also applied to improve the GA convergence speed. The experimental result shows the proposed method can obtain an accurate solution in a practical response time. © 2009 Wiley Periodicals, Inc. Electron Comm Jpn, 91(9): 1– 10, 2008; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10042 [ABSTRACT FROM AUTHOR]

Many kinds of evolutionary methods have been proposed. GA and GP in particular have demonstrated their effectiveness in various problems recently, and many systems have been proposed. One is Virus-Evolutionary Genetic Algorithm (VE-GA), and the other is Linear Genetic Programming in C (LGPC). The performance of each system has been reported. VE-GA is the coevolution system of host individuals and virus individuals. That can spread schema effectively among the host individuals by using virus infection and virus incorporation. LGPC implements the GP by representing the individuals to one dimension as if GA. LGPC can reduce a search cost of pointer and save machine memory, and can reduce the time to implement GP programs. We have proposed that a system introduce virus individuals in LGPC, and analyzed the performance of the system on two problems. Our system can spread schema among the population, and search solution effectively. The results of computer simulation show that this system can search for solution depending on LGPC applying problem's character compared with LGPC. © 2008 Wiley Periodicals, Inc. Electron Comm Jpn, 91(1): 32– 39, 2008; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.10030 [ABSTRACT FROM AUTHOR]