A novel rough neural network (RNN) structure and its application are proposed in this paper. We principally introduce its architecture and training algorithms: the genetic training algorithm (GA) and the tabu search training algorithm (TSA). We first compare RNN with the conventional NN trained by the BP algorithm in two-dimensional data classification. Then we compare RNN with NN by the same training algorithm (TSA) in functional approximation. Experiment results show that the proposed RNN is more effective than NN, not only in computation time but also in performance.

In this study, a hybrid genetic algorithm/neural network with fuzzy logic controller (NN-flcGA) is proposed to find the global optimum of reliability assignment/redundant allocation problems which should be simultaneously determined two different types of decision variables. Several researchers have obtained acceptable and satisfactory results using genetic algorithms for optimal reliability assignment/redundant allocation problems during the past decade. For large-size problems, however, genetic algorithms have to enumerate numerous feasible solutions due to the broad continuous search space. Recently, a hybridized GA combined with a neural network technique (NN-hGA) has been proposed to overcome this kind of difficulty. Unfortunately, it requires a high computational cost though NN-hGA leads to a robuster and steadier global optimum irrespective of the various initial conditions of the problems. The efficacy and efficiency of the NN-flcGA is demonstrated by comparing its results with those of other traditional methods in numerical experiments. The essential features of NN-flcGA namely, 1) its combination with a neural network (NN) technique to devise initial values for the GA, 2) its application of the concept of a fuzzy logic controller when tuning strategy GA parameters dynamically, and 3) its incorporation of the revised simplex search method, make it possible not only to improve the quality of solutions but also to reduce computational cost.

Spatial dynamic pattern formations or trails by organisms attract us, which remind us chaos and fractal. They seem to show the emergence of co-operation, job separation, or division of territories when genetic programming controls the reproduction, mutation, crossing over of the organisms. Recent research in social insect behavior suggests that swarm intelligence comes from pheromone or chemical trails, and models based on self-organization can help explain how colony-level behavior emerges out of interactions among individual insects. We try to explain the co-operative behaviors of social insect by means of density of organisms and their interaction with environment in simple simulations. We also study that MDL-based fitness evaluation is effective for improvement of generalization of genetic programming. At last, interspecific and intraspecific mathematical models are examined to expand our research into interspecific evolution.

In this paper, a hardware-oriented Genetic Algorithm (GA) was proposed in order to save the hardware resources and to reduce the execution time of GAP. Based on steady-state model among continuous generation model, the proposed GA used modified tournament selection, as well as special survival condition, with replaced whenever the offspring's fitness is better than worse-fit parent's. The proposed algorithm shows more than 30% in convergence speed over the conventional algorithm. Finally, by employing the efficient pipeline parallelization and handshaking protocol in proposed GAP, above 30% of the computation speed-up can be achieved over survival-based GA which runs one million crossovers per second (1 MHz), when device speed and size of application are taken into account on prototype. It would be used for high speed processing such of central processor of evolvable hardware, robot control and many optimization problems.

In this paper, genetic algorithms is employed to determine the shape of a conducting cylinder buried in a half-space. Assume that a conducting cylinder of unknown shape is buried in one half-space and scatters the field incident from another half-space where the scattered filed is measured. Based on the boundary condition and the measured scattered field, a set of nonlinear integral equations is derived and the imaging problem is reformulated into an optimization problem. The genetic algorithm is then employed to find out the nearly global extreme solution of the object function such that the shape of the conducting scatterer can be suitably reconstructed. In our study, even when the initial guess is far away from the exact one, the genetic algorithm can avoid the local extremes and converge to a reasonably good solution. In such cases, the gradient-based methods often get stuck in local extremes. Numerical results are presented and good reconstruction is obtained both with and without the additive Gaussian noise.

In order to maintain the diversity of structures in the population and prevent premature convergence, I have developed a new genetic algorithm called DCGA. In the experiments on many standard benchmark problems, DCGA showed good performances, whereas with harder problems, in some cases, the phenomena were observed that the search was stagnated at a local optimum despite that the diversity of the population is maintained. In this paper, I propose methods for escaping such phenomena and improving the performance by reinitializing the population, that is, a method called each-structure-based reinitializing method with a deterministic structure diverging procedure as a method for producing new structures and an adaptive improvement probability bound as a search termination criterion. The results of experiments demonstrate that DCGA becomes robust in harder problems by employing these proposed methods.

An optimal allocation model for the sub-processing-element (sub-PE) level redundancy is developed, which is solved by the genetic algorithms. In the allocation model, the average defect density D and the parameter δ are also considered in order to accurately analyze the element yield, where δ is defined as the ratio of the support circuit area to the total area of a PE. When the PE's area is imposed on the constraint, the optimal solutions of the model with different D and δ are calculated. The simulation results indicate that, for any fixed average defect density D, both the number of the optimal redundant sub-circuit added into a PE and the PE's yield decrease as δ increases. Moreover, for any fixed parameter δ, the number of the optimal redundant sub-circuit increases, while the optimal yield of the PE decreases, as D increases.

This letter presents an efficient graph-based evolutionary optimization technique, and its application to the transistor-level design of multiple-valued arithmetic circuits. The key idea is to introduce "circuit graphs with colored terminals" for modeling heterogeneous networks of various components. The potential of the proposed approach is demonstrated through experimental synthesis of a radix-4 signed-digit (SD) full adder circuit.

This paper proposes the Vector Evaluated GA-ICT (VEGA-ICT), a novel design method that employs the Genetic Algorithm (GA) to obtain the optimum antenna design. GA-ICT incorporates an arbitrary wire-grid model antenna to derive the optimum solution without any basic structure or limitation on the number of elements by merely optimizing an objective function. GA-ICT comprises the GA and an analysis method, the Improved Circuit Theory (ICT), with the following characteristics. (1) To achieve optimization of an arbitrary wire-grid model antenna without a basic antenna structure, the unknowns of the ICT are directly assigned to variables of the GA in the GA-ICT. (2) To achieve a variable number of elements, duplicate elements generated by using the same feasible region are deleted in the ICT. (3) To satisfy all complex design conditions, the GA-ICT generates an objective function using a weighting function generated based on electrical characteristics, antenna configuration, and size. (4) To overcome the difficulty of convergence caused by the nonlinearity of each term in the objective function, GA-ICT adopts a vector evaluation method. In this paper, the novel GA-ICT method is applied to downsize sector antennas. The calculation region in GA-ICT is reduced by adopting cylindrical coordinates and a periodic imaging structure. The GA-ICT achieves a 30% reduction in size compared to the previously reported small sector antenna, MS-MPYA, while retaining almost the same characteristics.

We propose the kernel-based pattern recognition hardware and its design methodology using the genetic algorithm. In the proposed design methodology, pattern data are transformed into the truth tables and the truth tables are evolved to represent kernels in the discrimination functions for pattern recognition. The evolved truth tables are then synthesized to logic circuits. Because of this data direct implementation approach, no floating point numerical circuits are required and the intrinsic parallelism in the pattern data set is embedded into the circuits. Consequently, high speed recognition systems can be realized with acceptable small circuit size. We have applied this methodology to the image recognition and the sonar spectrum recognition tasks, and implemented them onto the newly developed FPGA-based reconfigurable pattern recognition board. The developed system demonstrates higher recognition accuracy and much faster processing speed than the conventional approaches.

A plasma display panel (PDP) represents gray levels by the pulse number modulation technique that results in undesirable dynamic false contours on moving images. Among the various techniques proposed for the reduction of dynamic false contours, the optimization of the subfield pattern can be most easily implemented without the need for any additional dedicated hardware or software. In this paper, a systematic method for selecting the optimum subfield pattern is presented. In the proposed method, a subfield pattern that minimizes the quantitative measure of the dynamic false contour on the predefined test image is selected as the optimum pattern. The selection is made by repetitive calculations based on a genetic algorithm. Quantitative measure of the dynamic false contour calculated by simulation on the test image serves as a criterion for minimization by the genetic algorithm. In order to utilize the genetic algorithm, a structure of a string is proposed to satisfy the requirements for the subfield pattern. Also, three genetic operators for optimization, reproduction, crossover, and mutation, are specially designed for the selection of the optimum subfield pattern.

An automated analog circuit synthesis based on reuse of topological features of 'prototype circuits' is proposed. The prototype circuits are designed by humans and suggested to the synthesis system as hints of configurations of new analog circuits to be synthesized by the system. The connections of elements in analog circuits are not generally systematic, but they would have some similarities to a circuit which has similar behaviors or functionalities. In the proposed process, the information on circuit connections is stored as sub-circuits extracted from the prototype circuits. And then, genetic algorithm is used to search for an optimum combination of the sub-circuits that achieves the desired electronic specifications. The combinations of sub-circuits are performed with a novel technique where the terminals of the sub-circuits are shared. The capabilities of the proposed method are demonstrated through an example of the synthesis.

In this paper, we propose an approach to solve the power control issue in a DS-CDMA cellular system using genetic algorithms (GAs). The transmitter power control developed in this paper has been proven to be efficient to control co-channel interference, to increase bandwidth utilization and to balance the comprehensive services that are sharing among all the mobiles with attaining a common signal-to-interference ratio(SIR). Most of the previous studies have assumed that the transmitter power level is controlled in a constant domain under the assumption of uniform distribution of users in the coverage area or in a continuous domain. In this paper, the optimal centralized power control (CPC) vector is characterized and its optimal solution for CPC is presented using GAs in a large-scale DS-CDMA cellular system under the realistic context that means random allocation of active users in the entire coverage area. Emphasis is put on the balance of services and convergence rate by using GAs.

This paper presents an automatic synthesis method of active analog circuits that uses evolutionary search and employs some topological features of analog integrated circuits. Our system firstly generates a set of circuits at random, and then evolves their topologies and device sizing to fit an environment which is formed by the fitness function translated from the electrical specifications of the circuit. Therefore expert knowledge about circuit topologies and sizing are not needed. The capability of this method is demonstrated through experiments of automatic synthesis of CMOS operational amplifiers.

This paper deals with the control of chaotic dynamics by using the approximated system equations which are obtained by using the Genetic Programming (GP). Well known OGY method utilizes already existing unstable orbits embedded in the chaotic attractor, and use linearlization of system equations and small perturbation for control. However, in the OGY method we need transition time to attain the control, and the noise included in the linealization of equations moves the orbit into unstable region again. In this paper we propose a control method which utilize the estimated system equations obtained by the GP so that the direct nonlinear control is applicable to the unstable orbit at any time. In the GP, the system equations are represented by parse trees and the performance (fitness) of each individual is defined as the inversion of the root mean square error between the observed data and the output of the system equation. By selecting a pair of individuals having higher fitness, the crossover operation is applied to generate new individuals. In the simulation study, the method is applied at first to the artificially generated chaotic dynamics such as the Logistic map and the Henon map. The error of approximation is evaluated based upon the prediction error. The effect of noise included in the time series on the approximation is also discussed. In our control, since the system equations are estimated, we only need to change the input incrementally so that the system moves to the stable region. By assuming the targeted dynamic system f(x(t)) with input u(t)=0 is estimated by using the GP (denoted (x(t))), then we impose the input u(t) so that xf=(t+1)=(x(t))+u(t) where xf is the fixed point. Then, the next state x(t+1) of targeted dynamic system f(x(t)) is replaced by x(t+1)+u(t). The control method is applied to the approximation and control of chaotic dynamics generating various time series and even noisy time series by using one dimensional and higher dimensional system. As a result, if the noise level is relatively large, the method of the paper provides better control compared to conventional OGY method.

This paper deals with the approximation of multi-dimensional chaotic dynamics by using the multi-stage fuzzy inference system. The number of rules included in multi-stage fuzzy inference systems is remarkably smaller compared to conventional fuzzy inference systems where the number of rules are proportional to an exponential of the number of input variables. We also propose a method to optimize the shape of membership function and the appropriate selection of input variables based upon the genetic algorithm (GA). The method is applied to the approximation of typical multi-dimensional chaotic dynamics. By dividing the inference system into multiple stages, the total number of rules is sufficiently depressed compared to the single stage system. In each stage of inference only a portion of input variables are used as the input, and output of the stage is treated as an input to the next stage. To give better performance, the shape of the membership function of the inference rules is optimized by using the GA. Each individual corresponds to an inference system, and its fitness is defined by using the prediction error. Experimental results lead us to a relevant selection of the number of input variables and the number of stages by considering the computational cost and the requirement. Besides the GA in the optimization of membership function, we use the GA to determine the input variables and the number of input. The selection of input variable to each stage, and the number of stages are also discussed. The simulation study for multi-dimensional chaotic dynamics shows that the inference system gives better prediction compared to the prediction by the neural network.

A halftoning technique that uses a simple GA has proven to be very effective to generate high quality halftone images. Recently, the two major drawbacks of this conventional halftoning technique with GAs, i.e. it uses a substantial amount of computer memory and processing time, have been overcome by using an improved GA (GA-SRM) that applies genetic operators in parallel putting them in a cooperative-competitive stand with each other. The halftoning problem is a true multiobjective optimization problem. However, so far, the GA based halftoning techniques have treated the problem as a single objective optimization problem. In this work, the improved GA-SRM is extended to a multiobjective optimization GA to simultaneously generate halftone images with various combinations of gray level precision and spatial resolution. Simulation results verify that the proposed scheme can effectively generate several high quality images simultaneously in a single run reducing even further the overall processing time.

This paper proposes distributed evolutionary digital filters (EDFs) as an improved version of the original EDF. The EDF is an adaptive digital filter which is controlled by adaptive algorithm based on evolutionary computation. In the proposed method, a large population of the original EDF is divided into smaller subpopulations. Each sub-EDF has one subpopulation and executes the small-sized main loop of the original EDF. In addition, the distributed algorithm periodically selects promising individuals from each subpopulation. Then, they migrate to different subpopulations. Numerical examples show that the distributed EDF has a higher convergence rate and smaller steady-state value of the square error than the LMS adaptive digital filter, the adaptive digital filter based on the simple genetic algorithm and the original EDF.

The ultrasonic wave is widely used for acquiring perceptual information necessary for indoor/outdoor navigation of mobile robots, where the system is implemented as a sound navigation and ranging system (sonar). A robot equipped with multiple ultrasonic sonars is likely to exhibit undesirable operation due to erroneous measurements resulting from cross-talk among the sonars. Each sonar transmits and receives a pulse-modulated ultrasonic wave for measuring the range and identifying its own signal. We propose a technique for generating pulse patterns for multiple concurrently operated ultrasonic sonars. The approach considers pulse-pattern generation as a combinatorial optimization problem which can be solved by a genetic algorithm (GA). The aim is to acquire a pulse pattern satisfying certain conditions in order to avoid cross-talk or keep the probability of erroneous measurement caused by cross-talk low. We provide a method of genotype coding for the generation of the pulse pattern. Furthermore, in order to avoid a futile search encountered when the conventional technique is used, we propose an improved genotype coding technique that yields considerably different results from those of the conventional technique.

This paper presents a genetic algorithm for designing at minimum cost a two-connected network topology such that the shortest cycle (referred to as a ring) to which each edge belongs does not exceed a given maximum number of hops. The genetic algorithm introduces a solution representation in which constraints such as connectivity and ring constraints are easily encoded. Furthermore, a problem specific crossover operator that ensures solutions generated through genetic evolution are all feasible is also proposed. Hence, both checking of the constraints and repair mechanism can be avoided thus resulting in increased efficiency. Experimental evaluation shows the effectiveness of the proposed GA.