2006
DocumentsDate added
FTXI: Fault Tolerance XCS in Integer
Abstract: In the realm of data mining, several key issues exists in the traditional classification algorithms, such as low readability, large rule number, and low accuracy with information losing. In this paper, we propose a new classification methodology, called fault tolerance XCS in integer (FTXI), by extending XCS to handle conditions in integers and integrating the mechanism of fault tolerance in the context of data mining into the framework of XCS. We also design and generate appropriate artificial data sets for examine and verify the proposed method. Using the real world data as well, our experiments indicate that FTXI can provide the least rule number, obtain high prediction accuracy, and offer rule readability, compared to C4.5 and XCS in integer without fault tolerance.
Abstract: In the realm of data mining, several key issues exists in the traditional classification algorithms, such as low readability, large rule number, and low accuracy with information losing. In this paper, we propose a new classification methodology, called fault tolerance XCS in integer (FTXI), by extending XCS to handle conditions in integers and integrating the mechanism of fault tolerance in the context of data mining into the framework of XCS. We also design and generate appropriate artificial data sets for examine and verify the proposed method. Using the real world data as well, our experiments indicate that FTXI can provide the least rule number, obtain high prediction accuracy, and offer rule readability, compared to C4.5 and XCS in integer without fault tolerance.
A Hybrid Approach for Iterative Image Retrieval with Keywords and Visual Features
Abstract: QBK is an image search approach based on text description. The advantage of QBK is that it is based on semantics of mankind, and assisted by the matured text-based search technology. However, the disadvantage of QBK is that the result of image search is not affected by the content of the image itself. Besides, the text description does not represent the content of image fully.
CBIR is another image search approach which is based on the visual features of image itself. The advantage of CBIR is that the result of image search is all based on the content of the image and, it is objectively. The disadvantage of CBIR is that the basic technology is not matured enough. So the approach cannot imitate the recognition ability of human beings.
Our approach is the combination of QBK and CBIR which integrates the advantage of visual features and text description. This approach not only access the semantics, but also base on the content of image.
We extract the visual features with the method of CBIR from the result images of the QBK system. And then the images will be clustered by their visual features. Finally, users can iteratively search with keyword suggestions which are extracted from the description of clustered images.
Abstract: QBK is an image search approach based on text description. The advantage of QBK is that it is based on semantics of mankind, and assisted by the matured text-based search technology. However, the disadvantage of QBK is that the result of image search is not affected by the content of the image itself. Besides, the text description does not represent the content of image fully.
CBIR is another image search approach which is based on the visual features of image itself. The advantage of CBIR is that the result of image search is all based on the content of the image and, it is objectively. The disadvantage of CBIR is that the basic technology is not matured enough. So the approach cannot imitate the recognition ability of human beings.
Our approach is the combination of QBK and CBIR which integrates the advantage of visual features and text description. This approach not only access the semantics, but also base on the content of image.
We extract the visual features with the method of CBIR from the result images of the QBK system. And then the images will be clustered by their visual features. Finally, users can iteratively search with keyword suggestions which are extracted from the description of clustered images.
Evolutionary Interactive Music Composition
Abstract: This paper proposes and describes the CFE framework—Composition, Feedback, and Evolution—and presents an interactive music composition system. The system composes short, manageable pieces of music by interacting with users. The most important features of the system include creating customized music according to the user preference and the facilities specifically designed for producing massive music. We present the structure as well as the implementation of the system and the auxiliary functionalities that enhance the system. We also introduce the auto-feedback test with which we verify and evaluate the interactive music composition system, followed by the discussion and conclusions.
Abstract: This paper proposes and describes the CFE framework—Composition, Feedback, and Evolution—and presents an interactive music composition system. The system composes short, manageable pieces of music by interacting with users. The most important features of the system include creating customized music according to the user preference and the facilities specifically designed for producing massive music. We present the structure as well as the implementation of the system and the auxiliary functionalities that enhance the system. We also introduce the auto-feedback test with which we verify and evaluate the interactive music composition system, followed by the discussion and conclusions.
Introducing Recombination with Dynamic Linkage Discovery to Particle Swarm Optimization
Abstract: In this paper, we introduce the recombination operator with the technique of dynamic linkage discovery to particle swarm optimization (PSO) in order to improve the performance of PSO. Dynamic linkage discovery is a costless, effective linkage recognition technique adapting the linkage configuration by utilizing the natural selection without incorporating extra judging criteria irrelevant to the objective function. Furthermore, we employ a specific recombination operator to work with the building blocks identified by dynamic linkage discovery. Numerical experiments are conducted on a set of carefully designed benchmark functions and demonstrate good performance achieved by the proposed methodology.
Abstract: In this paper, we introduce the recombination operator with the technique of dynamic linkage discovery to particle swarm optimization (PSO) in order to improve the performance of PSO. Dynamic linkage discovery is a costless, effective linkage recognition technique adapting the linkage configuration by utilizing the natural selection without incorporating extra judging criteria irrelevant to the objective function. Furthermore, we employ a specific recombination operator to work with the building blocks identified by dynamic linkage discovery. Numerical experiments are conducted on a set of carefully designed benchmark functions and demonstrate good performance achieved by the proposed methodology.
Adaptive Discretization for Probabilistic Model Building Genetic Algorithms
Abstract: This paper proposes an adaptive discretization method, called Split-on-Demand (SoD), to enable the probabilistic model building genetic algorithm (PMBGA) to solve optimization problems in the continuous domain. The procedure, effect, and usage of SoD are described in detail. As an example, the integration of SoD and the extended compact genetic algorithm (ECGA), named real-coded ECGA (rECGA), is presented and numerically examined. The experimental results indicate that rECGA works well and SoD is effective. The behavior of SoD is analyzed and discussed, followed by the potential future work for SoD.
Abstract: This paper proposes an adaptive discretization method, called Split-on-Demand (SoD), to enable the probabilistic model building genetic algorithm (PMBGA) to solve optimization problems in the continuous domain. The procedure, effect, and usage of SoD are described in detail. As an example, the integration of SoD and the extended compact genetic algorithm (ECGA), named real-coded ECGA (rECGA), is presented and numerically examined. The experimental results indicate that rECGA works well and SoD is effective. The behavior of SoD is analyzed and discussed, followed by the potential future work for SoD.
iECGA: Integer Extended Compact Genetic Algorithm
Abstract: Extended compact genetic algorithm (ECGA) is an algorithm that can solve hard problems in the binary domain. ECGA is reliable and accurate because of the capability of detecting building blocks, but certain difficulties are encountered when we directly apply ECGA to problems in the integer domain. In this paper, we propose a new algorithm that extends ECGA, called integer extended compact genetic algorithm (iECGA). iECGA uses a modified probability model and inherits the capability of detecting building blocks from ECGA. iECGA is specifically designed for problems in the integer domain and can avoid the difficulties that ECGA encounters. In the experimental results, we show the performance comparisons between ECGA, iECGA, and a simple GA, and the results indicate that iECGA has good performances on problems in the integer domain.
Abstract: Extended compact genetic algorithm (ECGA) is an algorithm that can solve hard problems in the binary domain. ECGA is reliable and accurate because of the capability of detecting building blocks, but certain difficulties are encountered when we directly apply ECGA to problems in the integer domain. In this paper, we propose a new algorithm that extends ECGA, called integer extended compact genetic algorithm (iECGA). iECGA uses a modified probability model and inherits the capability of detecting building blocks from ECGA. iECGA is specifically designed for problems in the integer domain and can avoid the difficulties that ECGA encounters. In the experimental results, we show the performance comparisons between ECGA, iECGA, and a simple GA, and the results indicate that iECGA has good performances on problems in the integer domain.
Introducing Recombination with Dynamic Linkage Discovery to Particle Swarm Optimization
Abstract: There are two main objectives in this thesis. The first goal is to improve the performance of the particle swarm optimizer by incorporating linkage concept which is an essential mechanism in genetic algorithms. To achieve this purpose, we need to know the characteristics of the particle swarm optimizer and the genetic linkage problem. Through survey of the particle swarm optimization and the linkage problem, we then figure out how to introduce the linkage concept to particle swarm optimizer. Another goal is to address the linkage problem in real-parameter optimization problems. We have to study different linkage learning techniques, and understand the meaning of genetic linkage in real-parameter problems. After that, we design a novel linkage identification technique to achieve this objective.
In this thesis, the existence of genetic linkages in real-parameter optimization problem and that genetic linkages are dynamically changed through the search process are the primary assumptions. With these assumptions, we develop the dynamic linkage discovery technique to address the linkage problem. Moreover, a special recombination operator is designed to promote the cooperation of particle swarm optimizer and linkage identification technique. In the consequence, we introduce the recombination operator with the technique of dynamic linkage discovery to particle swarm optimization (PSO). Dynamic linkage discovery is a costless, effective linkage recognition technique adapting the linkage configuration by utilizing the natural selection without incorporating extra judging criteria irrelevant to the objective function. Furthermore, we employ a specific recombination operator to work with the building blocks identified by dynamic linkage discovery. The whole framework forms a new efficient search algorithm and is called PSO-RDL in this study. Numerical experiments are conducted on a set of carefully designed benchmark functions and demonstrate good performance achieved by the proposed methodology. Moreover, we also applied the proposed algorithm on the economic dispatch problem which is an essential topic in power control systems. The experimental results show that PSO-RDL can performs well both on numerical benchmark and real-world applications.
Abstract: There are two main objectives in this thesis. The first goal is to improve the performance of the particle swarm optimizer by incorporating linkage concept which is an essential mechanism in genetic algorithms. To achieve this purpose, we need to know the characteristics of the particle swarm optimizer and the genetic linkage problem. Through survey of the particle swarm optimization and the linkage problem, we then figure out how to introduce the linkage concept to particle swarm optimizer. Another goal is to address the linkage problem in real-parameter optimization problems. We have to study different linkage learning techniques, and understand the meaning of genetic linkage in real-parameter problems. After that, we design a novel linkage identification technique to achieve this objective.
In this thesis, the existence of genetic linkages in real-parameter optimization problem and that genetic linkages are dynamically changed through the search process are the primary assumptions. With these assumptions, we develop the dynamic linkage discovery technique to address the linkage problem. Moreover, a special recombination operator is designed to promote the cooperation of particle swarm optimizer and linkage identification technique. In the consequence, we introduce the recombination operator with the technique of dynamic linkage discovery to particle swarm optimization (PSO). Dynamic linkage discovery is a costless, effective linkage recognition technique adapting the linkage configuration by utilizing the natural selection without incorporating extra judging criteria irrelevant to the objective function. Furthermore, we employ a specific recombination operator to work with the building blocks identified by dynamic linkage discovery. The whole framework forms a new efficient search algorithm and is called PSO-RDL in this study. Numerical experiments are conducted on a set of carefully designed benchmark functions and demonstrate good performance achieved by the proposed methodology. Moreover, we also applied the proposed algorithm on the economic dispatch problem which is an essential topic in power control systems. The experimental results show that PSO-RDL can performs well both on numerical benchmark and real-world applications.
Particle Swarm Guided Evolution Strategy for Real-Parameter Optimization
Abstract: Evolutionary algorithms are stochastic search methods that mimic the metaphor of natural biological evolution. Among of these algorithms, evolution strategy (ES) and particle swarm optimization (PSO) are two of the most popular research topics. Both of ES and PSO are deal with real-parameter optimization problems but have different search behaviors. By observing the search behavior of ES and PSO, we find both of them have strength and weakness. In ES, the mutation operator lacks an explicit mechanism to guide the search into promising direction due to random variance. Moreover, there is no coordination in the movement of individuals within the search space. However, the powerful selection procedure allows solutions with superior characteristics to pass these from generation to generation. In the PSO, the search mechanism used the swarm cooperation concept. Each particle will move toward the direction which is expected to be good. The objective of this article is tried to combine ES and PSO at the concept level. We introduce the concept of swarm cooperation of PSO into the mutation operator of ES for reducing the disturbance of mutation in the mutation direction. We proposed a new mutation operator called guided mutation. Combining the guided mutation into the traditional ES framework and develop a new optimization algorithm called particle swarm guided evolution strategy. Numerical experiments are conducted on a set of carefully designed benchmark functions and demonstrate good performance achieved by the proposed methodology.
Abstract: Evolutionary algorithms are stochastic search methods that mimic the metaphor of natural biological evolution. Among of these algorithms, evolution strategy (ES) and particle swarm optimization (PSO) are two of the most popular research topics. Both of ES and PSO are deal with real-parameter optimization problems but have different search behaviors. By observing the search behavior of ES and PSO, we find both of them have strength and weakness. In ES, the mutation operator lacks an explicit mechanism to guide the search into promising direction due to random variance. Moreover, there is no coordination in the movement of individuals within the search space. However, the powerful selection procedure allows solutions with superior characteristics to pass these from generation to generation. In the PSO, the search mechanism used the swarm cooperation concept. Each particle will move toward the direction which is expected to be good. The objective of this article is tried to combine ES and PSO at the concept level. We introduce the concept of swarm cooperation of PSO into the mutation operator of ES for reducing the disturbance of mutation in the mutation direction. We proposed a new mutation operator called guided mutation. Combining the guided mutation into the traditional ES framework and develop a new optimization algorithm called particle swarm guided evolution strategy. Numerical experiments are conducted on a set of carefully designed benchmark functions and demonstrate good performance achieved by the proposed methodology.
