Shinn-Ying Ho

Design of accurate classifiers with a compact fuzzy-rule base using an evolutionary scatter partition of feature space

An evolutionary approach to designing accurate classifiers with a compact fuzzy-rule base using a scatter partition of feature space is proposed, in which all the elements of the fuzzy classifier design problem have been moved in parameters of a complex optimization problem. An intelligent genetic algorithm (IGA) is used to effectively solve the design problem of fuzzy classifiers with many tuning parameters. The merits of the proposed method are threefold: 1) the proposed method has high search ability to efficiently find fuzzy rule-based systems with high fitness values, 2) obtained fuzzy rules have high interpretability, and 3) obtained compact classifiers have high classification accuracy on unseen test patterns. The sensitivity of control parameters of the proposed method is empirically analyzed to show the robustness of the IGA-based method. The performance comparison and statistical analysis of experimental results using ten-fold cross validation show that the IGA-based method without heuristics is efficient in designing accurate and compact fuzzy classifiers using 11 well-known data sets with numerical attribute values.

Inheritable genetic algorithm for biobjective 0/1 combinatorial optimization problems and its applications

In this paper, we formulate a special type of multiobjective optimization problems, named biobjective 0/1 combinatorial optimization problem BOCOP, and propose an inheritable genetic algorithm IGA with orthogonal array crossover (OAX) to efficiently find a complete set of nondominated solutions to BOCOP. BOCOP with n binary variables has two incommensurable and often competing objectives: minimizing the sum r of values of all binary variables and optimizing the system performance. BOCOP is NP-hard having a finite number C(n, r) of feasible solutions for a limited number r. The merits of IGA are threefold as follows: 1) OAX with the systematic reasoning ability based on orthogonal experimental design can efficiently explore the search space of C(n, r); 2) IGA can efficiently search the space of C(n, r+/-1) by inheriting a good solution in the space of C(n, r); and 3) The single-objective IGA can economically obtain a complete set of high-quality nondominated solutions in a single run. Two applications of BOCOP are used to illustrate the effectiveness of the proposed algorithm: polygonal approximation problem (PAP) and the problem of editing a minimum reference set for nearest neighbor classification (MRSP). It is shown empirically that IGA is efficient in finding complete sets of nondominated solutions to PAP and MRSP, compared with some existing methods.

Design and Analysis of an Efficient Evolutionary Image Segmentation Algorithm

Evolutionary image segmentation algorithms have a number of advantages such as continuous contour, non-oversegmentation, and non-thresholds. However, most of the evolutionary image segmentation algorithms suffer from long computation time because the number of encoding parameters is large. In this paper, design and analysis of an efficient evolutionary image segmentation algorithm EISA are proposed. EISA uses a K-means algorithm to split an image into many homogeneous regions, and then uses an intelligent genetic algorithm IGA associated with an effective chromosome encoding method to merge the regions automatically such that the objective of the desired segmentation can be effectively achieved, where IGA is superior to conventional genetic algorithms in solving large parameter optimization problems. High performance of EISA is illustrated in terms of both the evaluation performance and computation time, compared with some current segmentation methods. It is empirically shown that EISA is robust and efficient using nature images with various characteristics.

A novel orthogonal simulated annealing algorithm for optimization of electromagnetic problems

We propose a novel orthogonal simulated annealing algorithm (OSA) for optimization of electromagnetic problems. The algorithm performs best when it employs an intelligent generation mechanism (IGM) based on orthogonal experimental design (OED). The OED-based IGM can efficiently generate a good candidate solution for the next step by using a systematic reasoning method instead of the conventional method of random perturbation. We show empirically that OSA is more efficient in solving parametric optimization problems and in designing optimal electromagnetic devices than some existing optimization methods using simulated annealing algorithms and genetic algorithms.

An efficient evolutionary image segmentation algorithm

In this paper, an efficient evolutionary image segmentation algorithm (EISA) is proposed. The existing evolutionary approach of image segmentation has the advantages over the other approaches such as continuous contour, non-oversegmentation, and non-thresholds, but suffers from long computation time. EISA uses a K-means algorithm to split an image into many homogeneous regions and then merges the split regions automatically using an evolutionary algorithm. The image segmentation problem is formulated as an optimization problem and the objective function is also given. EISA using a novel chromosome encoding method and a novel intelligent genetic algorithm makes the segmentation results robust and the computation time much shorter than the existing evolutionary image segmentation algorithms. Design and analysis of EISA are also presented. Experimental results of natural images with various degrees of noise demonstrate the effectiveness of EISA.

Design of nearest neighbor classifiers using an intelligent multi-objective evolutionary algorithm

The goal of designing optimal nearest neighbor classifiers is to maximize classification accuracy while minimizing the sizes of both reference and feature sets. A usual way is to adaptively weight the three objectives as an objective function and then use a single-objective optimization method for achieving this goal. This paper proposes a multi-objective approach to cope with the weight tuning problem for practitioners. A novel intelligent multi-objective evolutionary algorithm IMOEA is utilized to simultaneously edit compact reference and feature sets for nearest neighbor classification. Two comparison studies are designed to evaluate performance of the proposed approach. It is shown empirically that the IMOEA-designed classifiers have high classification accuracy and small sizes of reference and feature sets. Moreover, IMOEA can provide a set of good solutions for practitioners to choose from in a single run. The simulation results indicate that the IMOEA-based approach is an expedient method to design nearest neighbor classifiers, compared with an existing single- objective approach.

A Novel Multi-objective Orthogonal Simulated Annealing Algorithm for Solving Multi-objective Optimization Problems with a Large Number of Parameters

In this paper, a novel multi-objective orthogonal simulated annealing algorithm MOOSA using a generalized Pareto-based scale-independent fitness function and multi-objective intelligent generation mechanism (MOIGM) is proposed to efficiently solve multi-objective optimization problems with large parameters. Instead of generate-and-test methods, MOIGM makes use of a systematic reasoning ability of orthogonal experimental design to efficiently search for a set of Pareto solutions. It is shown empirically that MOOSA is comparable to some existing population-based algorithms in solving some multi-objective test functions with a large number of parameters.

An evolutionary approach for pose determination and interpretation of occluded articulated objects

This paper proposes a novel evolutionary approach to a parameter solving problem for handling occluded articulated objects with any number of internal parameters representing articulation. The parameter solving problem is formulated as a parameter optimization problem and an objective function is also given based on the line segment features in the Hough space. The proposed approach uses a novel intelligent genetic algorithm (IGA) superior to conventional GAs in solving large parameter optimization problems to simultaneously solve pose determination and interpretation problems and consequently has the capabilities of accurate partial matching and robust pose determination. Effectiveness of the proposed IGA-based method is demonstrated by applying it to fitting a simplified artificial articulated model of a human body to monocular clutter images.

A Novel Orthogonal Simulated Annealing Algorithm for Optimization of Electromagnetic Problems

We propose a novel orthogonal simulated annealing algorithm (OSA) for optimization of electromagnetic problems. The algorithm performs best when it employs an intelligent generation mechanism (IGM) based on orthogonal experimental design (OED). The OED-based IGM can efficiently generate a good candidate solution for the next step by using a systematic reasoning method instead of the conventional method of random perturbation. We show empirically that OSA is more efficient in solving parametric optimization problems and in designing optimal electromagnetic devices than some existing optimization methods using simulated annealing algorithms and genetic algorithms.

Design of high performance fuzzy controllers using flexible parameterized membership functions and intelligent genetic algorithms

This paper proposes a method for designing high performance fuzzy controllers with a compact rule system. The method is mainly derived from flexible parameterized membership functions (FPMFs) and a novel intelligent genetic algorithm (IGA). Each FPMF consists of flexible trapezoidal fuzzy sets and the fuzzy set is encoded by five parameters. Furthermore, the membership functions and fuzzy rules are simultaneously determined by effectively incorporating all the system parameters into chromosomes. Therefore, the optimal design of fuzzy controllers is formulated as a large parameter optimization problem, which can be effectively solved by IGA. The proposed method is demonstrated by two well-known problems, truck backing and cart centering problems. It is shown empirically that the performance of the proposed method is superior to those of existing methods in terms of the numbers of time steps and fuzzy rules.