This paper addresses the issue of reconfiguring multi-layer neural networks implemented in single or multiple VLSI chips. The ability to adaptively reconfigure network configuration for a given application, considering the presence of faulty neurons, is a very valuable feature in a large scale neural network. In addition, it has become necessary to achieve systems that can automatically reconfigure a network and acquire optimal weights without any help from host computers. However, self-reconfigurable architectures for neural networks have not been studied sufficiently. In this paper, we propose an architecture for a self-reconfigurable multi-layer neural network employing both reconfiguration with spare neurons and weight training by GAs. This proposal offers the combined advantages of low hardware overhead for adding spare neurons and fast weight training time. To show the possibility of self-reconfigurable neural networks, the prototype system has been implemented on a field programmable gate array.

The alignment of biological sequences is a crucial tool in molecular biology and genome analysis. A wide variety of approaches has been proposed for multiple sequence alignment problem; however, some of them need prerequisites to help find the best alignment or some of them may suffer from the drawbacks of complexity and memory requirement so they can be only applied to cases with a limited number of sequences. In this paper, we view the multiple sequence alignment problem as an optimization problem and propose a heuristic-based genetic algorithm (GA) approach to solve it. The heuristic/GA hybrid yields better results than other well-known packages do. Experimental results are presented to illustrate the feasibility of the proposed approach.

A new scheme for watermarking based on vector quantization (VQ) over a binary symmetric channel is proposed. By optimizing VQ indices with genetic algorithm, simulation results not only demonstrate effective transmission of watermarked image, but also reveal the robustness of the extracted watermark.

This paper proposes the use of the ratio of wavelet extrema numbers taken from the horizontal and vertical counts respectively as a texture feature, which is called aspect ratio of extrema number (AREN). We formulate the classification problem upon natural and synthesized texture images as an optimization problem and develop a coevolving approach to select both scalar wavelet and multiwavelet feature spaces of greater discriminatory power. Sequential searches and genetic algorithms (GAs) are comparatively investigated. The experiments using wavelet packet decompositions with the innovative packet-tree selection scheme ascertain that the classification accuracy of coevolutionary genetic algorithms (CGAs) is acceptable enough.

The number and location of the inventory centers play an important role in the material distribution process since residents and inventory centers may be in dispersed regions. In this paper, we view the problem of finding the better locations for the inventory centers as an optimization problem, and propose a nested genetic algorithm (NGA) approach to design an optimal material distribution system. We demonstrate the feasibility of the proposed approach by numerical experiments.

Simple genetic algorithm (SGA) is a population-based optimization method based on the Darwinian natural selection. The theoretical foundations of SGA are the Schema Theorem and the Building Block Hypothesis. Although SGA does well in many applications as an optimization method, it still does not guarantee the convergence of a global optimum in GA-hard problems and deceptive problems. As an alternative schema, therefore, there is a growing interest in a co-evolutionary system where two populations constantly interact and cooperate each other. In this paper we propose a schema co-evolutionary algorithm (SCEA) and show why the SCEA works better than SGA in terms of an extended schema theorem. The experimental analyses using the Walsh-Schema Transform show that the SCEA works well in GA-hard problems including deceptive problems.

Genetic algorithm (GA) is a widely used numerical technique to simplify some analytical solutions in electromagnetic theory. Genetic algorithms can be combined with the geometric optics method to tackle electromagnetic scattering problems. This paper presents an extrapolation procedure, which derived, as a first step, a functional representation of the radar cross section (RCS) of three different dielectric objects that was computed via the Mie solution or the method of moments (MOM). An algorithm was employed to fit the scattering characteristics of dielectric objects at high frequencies.

In this paper, an efficient optimization algorithm for solving the inverse problem of a two-dimensional lossless homogeneous dielectric object is investigated. A lossless homogeneous dielectric cylinder of unknown permittivity scatters the incident wave in free space and the scattered fields are recorded. Based on the boundary condition and the incident field, a set of nonlinear surface integral equation is derived. The imaging problem is reformulated into optimization problem and the steady-state genetic algorithm is employed to reconstruct the shape and the dielectric constant of the object. Numerical results show that the permittivity of the cylinders can be successfully reconstructed even when the permittivity is fairly large. The effect of random noise on imaging reconstruction is also investigated.

In this paper an algorithm based on Ant Colony Optimization techniques called Ants on a Tree (AOT) is introduced. This algorithm can integrate many algorithms together to solve a single problem. The strength of AOT is demonstrated by solving a High-Level Synthesis problem. A High-Level Synthesis problem consists of many design steps and many algorithms to solve each of them. AOT can easily integrate these algorithms to limit the search space and use them as heuristic weights to guide the search. During the search, AOT generates a dynamic decision tree. A boosting technique similar to branch and bound algorithms is applied to guide the search in the decision tree. The storage explosion problem is eliminated by the evaporation of pheromone trail generated by ants, the inherent property of our search algorithm.

We propose a frequency stabilization system for laser diodes (LDs), in which the electrical feedback loop response can be determined using an on-line genetic algorithm (GA) so as to attain lower LD frequency noise power within the specific Fourier frequency range of interest. At the initial stage of the stabilization, the feedback loop response has been controlled through GA, manipulating the proportional gain, integration time, and derivative time of conventional analog PID controller. Individuals having 12-bit chromosomes encoded by combinations of PID parameters have converged evolutionarily toward an optimal solution providing a suitable feedback loop response. A fitness function has been calculated for each individual in real time based on the power spectral density (PSD) of the frequency noise. The performance of this system has been tested by stabilizing a 50 mW visible LD. Long-term (τ > 0.01 s) frequency stability and its repeatability have been improved.

NK-Landscapes are stochastically generated fitness functions on bit strings, parameterized with N bits and K epistatic interactions between bits. The term epistasis describes nonlinearities in fitness functions due to changes in the values of interacting bits. Empirical studies have shown that the overall performance of random bit climbers on NK-Landscapes is superior to the performance of some simple and enhanced genetic algorithms (GAs). Analytical studies have also lead to suggest that NK-Landscapes may not be appropriate for testing the performance of GAs. In this work we study the effect of selection, drift, mutation, and recombination on NK-Landscapes for N = 96. We take a model of generational parallel varying mutation GA (GA-SRM) and switch on and off its major components to emphasize each of the four processes mentioned above. We observe that using an appropriate selection pressure and postponing drift make GAs quite robust on NK-Landscapes; different to previous studies, even simple GAs with these two features perform better than a random bit climber (RBC+) for a broad range of classes of problems (K 4). We also observe that the interaction of parallel varying mutation with crossover improves further the reliability of the GA, especially for 12 < K < 32. Contrary to intuition, we find that for small K a mutation only evolutionary algorithm (EA) is very effective and crossover may be omitted; but the relative importance of crossover interacting with varying mutation increases with K performing better than mutation alone (K > 12). This work indicates that NK-Landscapes are useful for testing each one of the major processes involved in a GA and for assessing the overall behavior of a GA on complex non-linear problems. This study also gives valuable guidance to practitioners applying GAs to real world problems of how to configure the GA to achieve better results as the non-linearity and complexity of the problem increases.

In this paper, a spatio-temporal error concealment method of transmission errors for improving visual quality over the wireless channel is proposed, which makes use of geometric information extracted from the surrounding blocks. The geometric information is an isophote that is curves of constant intensity of image. To improve visual quality during video communication, the proposed method smoothly connects the isophotes disconnected due to transmission error using a genetic algorithm (GA) with an isophote constraint. In the proposed method, the error concealment problem is modeled as an optimization problem, which in our case, is solved by a cost function with isophotes constraint that is minimized using a GA. Experimental results shows more visually realistic than other error concealment methods.

This paper presents the design consideration of a polarization-transformation transmission filter, which is composed of a multilayered chiral slab. The optimal material parameters and thickness of each layer of the slab can be determined by using a genetic algorithm (GA). Substituting the constitutive relations for a chiral medium into Maxwell's equations, the electromagnetic field in the medium is obtained. A chain-matrix formulation is used to derive the relationship between the components of the incident, the reflected, and the transmitted electric fields. The cross- and co-polarized powers carried by the transmitted and reflected waves are represented in terms of their electric field components. The procedure proposed for the design of a polarization-transformation filter is divided into two stages. An ordinary filter without polarization-transformation and a polarization-transformation filter for the transmitted wave are designed with a multilayered non-chiral slab and a multilayered chiral slab at the first and the second stages, respectively. According to the specifications of the filters, two functionals are defined with the transmitted and reflected powers. Thus the optimal design of a polarization-transformation filter with the multilayered chiral slab is reduced to an optimization problem where the material parameters and thickness of each chiral layer are found by maximizing the functionals. Applying the GA to the maximization of the functionals, one can obtain the optimal material parameters and thicknesses of the multilayered chiral slab. Numerical results are presented to confirm the effectiveness of the two-stage design procedure. For three types of multilayered chiral slabs, optimal values of refractive indices, thicknesses, and chiral admittances are obtained. It is seen from the numerical results that the proposed procedure is very effective in the optimal design of polarization-transformation filters for the transmitted wave.

A new approach to correcting the array amplitude failure by a combination of B-spline techniques and genetic algorithms is proposed. Some array elements indicate the control knots for a B-spline curve by their nominal positions and amplitudes; others distribute the excitation amplitudes according to the sampling points on the curve. The inherent smoothness of the B-spline curves reduce the effect of excessive coupling between adjacent elements. Genetic algorithms are used to search for a quasi-optimized B-spline curve to produce the ultimate amplitude distribution for correcting the array failure. To demonstrate the method's effectiveness, simulation results for correcting failures with three- and four-element failures are presented.

Multiple access interference and near-far effect cause the performance of the conventional single user detector in DS/CDMA systems to degrade. Due to high complexity of the optimum multiuser detector, suboptimal multiuser detectors with less complexity and reasonable performance have received considerable attention. In this paper we apply the classic and a new modified genetic algorithm for multiuser detection of DS/CDMA signals. It is shown that the classic genetic algorithm (GA) reaches an error floor at high signal to noise ratios (SNR) while the performance of proposed modified GA is much better than the classic one and is comparable to the optimum detector with much less complexity. The results hold true for AWGN and fading channels. We also describe another GA called as meta GA to find the optimum parameters of the modified GA. We compare the performance of proposed method with the other detectors used in CDMA.

Many fuzzy traffic controllers adjust the extension time of the green phase with the fuzzy input variables, arrival and queue. However, in our experiments, we found that the two input variables are not sufficient for an intersection where traffic flow rates change and thus, in this paper, traffic volume is used as an additional variable. Traffic volume is defined as the number of vehicles entering an intersection every second. In designing a fuzzy traffic controller, an ad-hoc approach is usually used to find membership functions and fuzzy control rules showing good performance. That is, initial ones are generated by human operators and modified many times based on the results of simulation. To partially overcome the limitations of the ad-hoc approach, we use genetic algorithms to automatically determine the membership functions for terms of each fuzzy variable when fuzzy control rules are given by hand. The experimental results indicate that a fuzzy logic controller with volume variable outperforms conventional ones with no volume variable in terms of the average delay and the average velocity. Also, the controller shows better performance when membership functions generated by a genetic algorithms instead of ones generated by hand are used.

This paper presents a new reconfigurable hardware paradigm, called evolvable hardware (EHW), and its application to the biomedical engineering problem of an artificial hand controller. Evolvable hardware is based on the idea of combining a reconfigurable hardware device with an artificial intelligence robust search technique called genetic algorithms (GAs) to execute reconfiguration autonomously. The first version of the EHW chip was designed in 1998, and this paper describes the latest improvements to the EHW chip, as well as outlining its architecture and the hardware implementation of the GA operations. Execution speed for genetic operations is shown to be about 38.7 times faster with the hardware implementation than with software program running on an AMD Athlon processor (1.2GHz). As an application of the EHW chip, this paper introduces a controller for a multi-functional prosthetic-hand, and presents experimental data in which a practical myoelectric pattern classification rate of 97.8% was achieved through the application of the EHW chip.

In this letter, base station placement is automatically determined from pre-defined candidate sites using a genetic approach, and the transmit power is obtained taking the interference situation into account in cases of interference-dominant systems. In order to apply a genetic algorithm to the base station placement problem, a real-valued representation scheme is proposed. Corresponding operators such as crossover and mutation are also introduced. The proposed algorithm is applied to an inhomogeneous traffic density environment, where a base station's coverage may be limited by offered traffic loads. An objective function is designed for performing the cell planning in a coverage- and cost-effective manner.

The wavelet transform (WT) has recently emerged as a powerful tool for image compression. In this paper, a new image compression technique combining the genetic algorithm (GA) and grey-based competitive learning network (GCLN) in the wavelet transform domain is proposed. In the GCLN, the grey theory is applied to a two-layer modified competitive learning network in order to generate optimal solution for VQ. In accordance with the degree of similarity measure between training vectors and codevectors, the grey relational analysis is used to measure the relationship degree among them. The GA is used in an attempt to optimize a specified objective function related to vector quantizer design. The physical processes of competition, selection and reproduction operating in populations are adopted in combination with GCLN to produce a superior genetic grey-based competitive learning network (GGCLN) for codebook design in image compression. The experimental results show that a promising codebook can be obtained using the proposed GGCLN and GGCLN with wavelet decomposition.

An automated synthesis for analog computational circuits in transistor-level configuration is presented. A cell-based structure is introduced to place moderate constraints on the MOSFET circuit topology. Even though each cell has a simple structure that consists of one current path with four transistors, common analog building blocks can be implemented using combinations of the cells. A genetic algorithm is applied to search circuit topologies and transistor sizes that satisfy given specifications. Synthesis capabilities are demonstrated through examples of three types of computational circuits; absolute value, squaring, and cubing functions by using computer simulations and real hardware.