Feature selection is an important technique of data processing in the field of machine learning and data mining. Its goal is to select the feature subset with the maximum classification accuracy and the minimum number...
详细信息
Feature selection is an important technique of data processing in the field of machine learning and data mining. Its goal is to select the feature subset with the maximum classification accuracy and the minimum number. Using the particleswarm algorithm to find the optimal sunset in the high-dimensional data set is faced with the problems of falling into the local optimum and expensive calculation, resulting in a decrease in classification accuracy. Focused on these, this paper proposes a hybrid simplified PSO-based feature selection algorithm with the elite strategy (HECSPSO). It has the following improvements: (1) In the stage of population initialization, according to the separation performance of features, the conditional separation probability (S_probability) of features is redefined, on the basis of which a new population initialization strategy is proposed. (2) In order to further improve the convergence speed of the algorithm, this paper proposes the addition and deletion criterion of maximum separation-minimum redundancy according to the separation and redundancy of features, which is called elite strategy. (3) In order to simplify the complexity of the model, a simplified particle swarm optimization algorithm is proposed. The evolution process is controlled only by the position, which simplifies the iterative process of particleswarm and avoids the problems of slow convergence and low precision caused by particle velocity. (4) In order to avoid the algorithm falling into local optimum, the chaotic mechanism is used as the local search operator near the known solutions. In order to make a comprehensive evaluation, the proposed method is compared with other algorithms based on particleswarmoptimization. On the 16 data sets of UCI (University of California Irvine Machine Learning Repository), these methods are compared and evaluated in three aspects: the classification accuracy, the selected feature subset size, and the number of iterations for the algorit
particleswarmoptimization (PSO) has been widely used in various optimization fields because of its easy implementation and high efficiency. However, it suffers from some limitations like slow convergence and prematu...
详细信息
particleswarmoptimization (PSO) has been widely used in various optimization fields because of its easy implementation and high efficiency. However, it suffers from some limitations like slow convergence and premature convergence when solving high-dimensional optimization problems. This paper attempts to address these open issues. Firstly, a new method of parameter adjustment named piecewise nonlinear acceleration coefficients is introduced to the simplified particle swarm optimization algorithm (SPSO), and an improved algorithm called piecewise-nonlinear-acceleration-coefficients-based SPSO (P-SPSO) is proposed. Then, a mean differential mutation strategy is developed for the update mechanism of P-SPSO, and another improved algorithm named mean-differential-mutation-strategy embedded P-SPSO (MP-SPSO) is proposed. To validate the performance of the proposed algorithms, four different sets of experiments are carried out in this paper. The results show that, 1) the proposed P-SPSO can get better solutions than other four classic improved SPSO with different acceleration coefficients, 2) the proposed MP-SPSO algorithm shows better optimization performance than P-SPSO and mean-differential-mutation-strategy-based SPSO (M-SPSO), 3) the proposed MP-SPSO is clearly seen to be more successful than other eight well-known PSO variants, 4) compared to other nine intelligent optimization algorithms, MP-SPSO achieves better performance in terms of solution quality and robustness. Moreover, the proposed MP-SPSO algorithm is successfully applied to a real constrained engineering problem and provides better solutions than other methods.
In this paper, a simplified particle swarm optimization algorithm based on backtracking search (BSAPSO) is proposed to solve the problems of weak global search ability, easy to be trapped into local optimal solution a...
详细信息
ISBN:
(数字)9781728157122
ISBN:
(纸本)9781728157122
In this paper, a simplified particle swarm optimization algorithm based on backtracking search (BSAPSO) is proposed to solve the problems of weak global search ability, easy to be trapped into local optimal solution and difficult to obtain optimal solution in complex problems. Firstly, the velocity term of the standard particleswarmoptimization is eliminated, and the exponential random term is introduced to replace the optimal solution of individual particles to simplify the particle iteration process. Then, the special historical backtracking of Backtracking Search algorithm (BSA) is used to increase the population diversity of particleswarmoptimization without increasing the complexity of the algorithm. At the same time, the cross-mutation mechanism is introduced to enhance the global search ability of the algorithm. Finally, the validity of the improved algorithm is verified from three aspects: the accuracy of the algorithm, the convergence speed and the statistical test.
Graphical User Interface (GUI) is the outer skin of programs that facilitate the interaction between the user and different type of computing devices. It is been used in different aspects ranging from normal computers...
详细信息
Graphical User Interface (GUI) is the outer skin of programs that facilitate the interaction between the user and different type of computing devices. It is been used in different aspects ranging from normal computers, mobile device, to even very small device nowadays like watches. This interaction uses different tools and programming objects like images, text, buttons, checkboxes, etc. With this emergence of different types of GUIs, they become an essential component to be tested (if available in the software) to ensure that the software meets the required quality by the user. In contrast to non-functional testing, function testing of GUI insures a proper interaction between the user and the application interface without dealing with the coding internals. In this paper, a strategy for GUI functional testing using simplifiedswarmoptimization (SSO) is proposed. The SSO is used to generate an optimized test suite with the help of Event-Interaction Graph (EIG). The proposed strategy also manages and repairs the test suites by deleting the unnecessary event sequences that are not applicable. The proposed generation algorithm based on SSO has proved its effectiveness by evaluating it against other algorithms. In addition, the strategy is applied on a standard case study and proved its applicability in reality. Copyright (C) 2014, Karabuk University. Production and hosting by Elsevier B.V. All rights reserved.
暂无评论