Sigeru Omatu

Web page feature selection and classification using neural networks

Automatic categorization is the only viable method to deal with the scaling problem of the World Wide Web (WWW). In this paper, we propose a news web page classification method (WPCM). The WPCM uses a neural network with inputs obtained by both the principal components and class profile-based features. Each news web page is represented by the term-weighting scheme. As the number of unique words in the collection set is big, the principal component analysis (PCA) has been used to select the most relevant features for the classification. Then the final output of the PCA is combined with the feature vectors from the class-profile which contains the most regular words in each class. We have manually selected the most regular words that exist in each class and weighted them using an entropy weighting scheme. The fixed number of regular words from each class will be used as a feature vectors together with the reduced principal components from the PCA. These feature vectors are then used as the input to the neural networks for classification. The experimental evaluation demonstrates that the WPCM method provides acceptable classification accuracy with the sports news datasets.

High speed paper currency recognition by neural networks

In this paper a new technique is proposed to improve the recognition ability and the transaction speed to classify the Japanese and US paper currency. Two types of data sets, time series data and Fourier power spectra, are used in this study. In both cases, they are directly used as inputs to the neural network. Furthermore, we also refer a new evaluation method of recognition ability. Meanwhile, a technique is proposed to reduce the input scale of the neural network without preventing the growth of recognition. This technique uses only a subset of the original data set which is obtained using random masks. The recognition ability of using large data set and a reduced data set are discussed. In addition to that the results of using a reduced data set of the Fourier power spectra and the time series data are compared.

Efficient Genetic Algorithms Using Simple Genes Exchange LocalSearch Policy for the Quadratic Assignment Problem

In this paper, we describe an approach for solving the quadratic assignment problem (QAP) that is based on genetic algorithms (GA). It will be shown that a standard canonical GA (SGA), which involves genetic operators of selection, reproduction, crossover, and mutation, tends to fall short of the desired performance expected of a search algorithm. The performance deteriorates significantly as the size of the problem increases. To address this syndrome, it is common for GA-based techniques to be embedded with deterministic local search procedures. It is proposed that the local search should involve simple procedure of genome reordering that should not be too complex. More importantly, from a computational point of view, the local search should not carry with it the full cost of evaluating a chromosome after each move in the localized landscape. Results of simulation on several difficult QAP benchmarks showed the effectiveness of our approaches.

Timetable planning using the constraint-based reasoning

Abstract College timetabling is a combinatoric and dynamic problem. In order to solve this problem, the approach must be efficient, flexible, portable, and adaptable. This paper describes a solution procedure based on a constraint-based reasoning technique implemented in an object-oriented approach. The problem is formulated as a constraint satisfaction model and it is then solved using the proposed algorithm. The algorithm is tested using real data from one of the colleges offering professional courses. The results show that a 18-weeks timetable for 1673 subject sections, 10 rooms, 21 lecturers can be solved in less than 33 minutes as compared to several weeks if it is to be solved manually. Since it is implemented using the object-oriented approach, the proposed system can also be modified and easily adapted to support changes. Scope and purpose Timetable planning is an activity of assigning subjects to time and space such that all constraints are satisfied. It can be categorized into several types and the most common types are academic (in universities, colleges, and schools), airline, and nurse timetablings. Constraint-based reasoning is a problem-solving technique that combines logic programming and constraint-solving technique based on an arc-consistency algorithm. It has been used to solve constraint satisfaction problems. The purpose of the paper is to present the modeling process of the college timetabling as a constraint satisfaction problem. The model is then solved using the constraint-based reasoning approach.

Ship maintenance scheduling by genetic algorithm and constraint-based reasoning

Ship maintenance scheduling is a process to decide start times of maintenance activities that satisfy all precedence and resource constraints and optimize the ship availability. In this paper, ship maintenance scheduling is modelled as a constraint satisfaction problem (CSP). The variables of CSP are the start times and its domain values are the start and horizon of the schedule. To solve the ship maintenance scheduling problem in the Royal Malaysian Navy, we have adopted a constraint-based reasoning (CBR) which requires start times of the first activities of maintenance cycles to solve the problem by the CBR. Thus, we adopt a genetic algorithm (GA) to find the start times of the first activities. The simulation results showed the effectiveness of the present hybrid algorithm.

VEP optimal channel selection using genetic algorithm for neural network classification of alcoholics

In this letter, neural networks (NNs) classify alcoholics and nonalcoholics using features extracted from visual evoked potential (VEP). A genetic algorithm (GA) is used to select the minimum number of channels that maximize classification performance. GA population fitness is evaluated using fuzzy ARTMAP (FA) NN, instead of the widely used multilayer perceptron (MLP). MLP, despite its effective classification, requires long training time (on the order of 10(3) times compared to FA). This causes it to be unsuitable to be used with GA, especially for on-line training. It is shown empirically that the optimal channel configuration selected by the proposed method is unbiased, i.e., it is optimal not only for FA but also for MLP classification. Therefore, it is proposed that for future experiments, these optimal channels could be considered for applications that involve classification of alcoholics.

Incorporating constraint propagation in genetic algorithm for university timetable planning

Abstract Timetable planning can be modelled as a constraint-satisfaction problem, and may be solved by various approaches, including genetic algorithms. An optimal solution for a timetable planning problem is difficult to find using genetic algorithms, due to the ambiguity in deciding the fitness function. Various approaches aimed at finding optimal solutions to constraint-satisfaction problems by genetic algorithms have been proposed, but most of these approaches are problem-dependent and hence are difficult to apply to real-world problems. In this paper, a hybrid algorithm consisting of a genetic algorithm and constraint-based reasoning is proposed to find a feasible and near-optimal solution. The proposed algorithm was tested by using real data for university timetable planning, and this approach can be applied to most constraint-satisfaction problems.

Extensive Testing of a Hybrid Genetic Algorithm for Solving Quadratic Assignment Problems

A robust search algorithm should ideally exhibit reasonable performance on a diverse and varied set of problems. In an earlier paper Lim et al. (Computational Optimization and Applications, vol. 15, no. 3, 2000), we outlined a class of hybrid genetic algorithms based on the k-gene exchange local search for solving the quadratic assignment problem (QAP). We follow up on our development of the algorithms by reporting in this paper the results of comprehensive testing of the hybrid genetic algorithms (GA) in solving QAP. Over a hundred instances of QAP benchmarks were tested using a standard set of parameters setting and the results are presented along with the results obtained using simple GA for comparisons. Results of our testing on all the benchmarks show that the hybrid GA can obtain good quality solutions of within 2.5% above the best-known solution for 98% of the instances of QAP benchmarks tested. The computation time is also reasonable. For all the instances tested, all except for one require computation time not exceeding one hour. The results will serve as a useful baseline for performance comparison against other algorithms using the QAP benchmarks as a basis for testing.

Self-tuning neuro-PID control and applications

In this paper, we propose a method to use the neural networks to tune the PID (proportional plus integral plus derivative) gains such that human operators tune the gains adaptively according to the environmental condition and systems specification. The tuning method is based on the error backpropagation method and hence it may be trapped in a local minimum. In order to avoid the local minimum problem, we use the genetic algorithm to find the initial values of the connection weights of the neural network and initial values of PID gains. The experimental results show the effectiveness of the present approach.

A combined negative selection algorithm-particle swarm optimization for an email spam detection system

Abstract Email is a convenient means of communication throughout the entire world today. The increased popularity of email spam in both text and images requires a real-time protection mechanism for the media flow. The previous approach has been limited by the adaptive nature of unsolicited email spam. This research introduces an email detection system that is designed based on an improvement in the negative selection algorithm. Furthermore, particle swarm optimization (PSO) was implemented to improve the random detector generation in the negative selection algorithm (NSA). The algorithm generates detectors in the random detector generation phase of the negative selection algorithm. The combined NSA–PSO uses a local outlier factor (LOF) as the fitness function for the detector generation. The detector generation process is terminated when the expected spam coverage is reached. A distance measure and a threshold value are employed to enhance the distinctiveness between the non-spam and spam detectors after the detector generation. The implementation and evaluation of the models are analyzed. The results show that the accuracy of the proposed NSA–PSO model is better than the accuracy of the standard NSA model. The proposed model with the best accuracy is further used to differentiate between spam and non-spam in a network that is developed based on a client–server network for spam detection.