PSO-Based Hyper-Parameters Selection for LS-SVM Classifiers

Guo, Xuexin; Wu, Chun; Wang, C. Y.

doi:10.1007/11893257_124

Cited by 20 publications

(23 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…x is the ith input features vector of d-dimension and i y is the class label of i x , which is either + 1 or -1 (Guo et al, 2006;Silver et al, 2006). In the feature space, the classification function of LS-SVM (Suykens et al, 2002;Guo et al, 2006) can be described as…”

Section: Least Square Support Vector Machine (Ls-svm) For Classificationmentioning

confidence: 99%

EEG signal classification based on simple random sampling technique with least square support vector machine

Siuly

Wen

2011

IJBET

View full text Add to dashboard Cite

This paper proposes a new approach based on Simple Random Sampling (SRS) technique with Least Square Support Vector Machine (LS-SVM) to classify two-class of electroencephalogram (EEG) signals.The experiments are carried out on two EEG databases and a synthetic Ripley dataset. All two-class pairs are tested and our proposed approach obtains a 95.58% average classification accuracy for the EEG epileptic database, 98.73% for the mental imagery tasks EEG database and 100% for Ripley data. We compare our method with two most recent methods for the epileptic database. Experimental results demonstrate that the proposed method is more promising than previously reported classification techniques.

show abstract

Section: Least Square Support Vector Machine (Ls-svm) For Classificationmentioning

confidence: 99%

EEG signal classification based on simple random sampling technique with least square support vector machine

Siuly

Wen

2011

IJBET

View full text Add to dashboard Cite

show abstract

“…There are two key factors to determine the optimized hyper-parameters using QPSO: one is how to represent the hyper-parameters as the particle's position, namely how to encode [10,11]. Another is how to define the fitness function, which evaluates the goodness of a particle.…”

Section: Hyper-parameters Selection Based Qpsomentioning

confidence: 99%

Model Selection of Least Squares Support Vector Regression Using Quantum-Behaved Particle Swarm Optimization Algorithm

Zhou²,

Liu

2009

2009 International Workshop on Intelligent Systems and Applications

View full text Add to dashboard Cite

The selection for hyper-parameters is difficult and important to the performance of Least Squares SupportVector Machines (LS-SVM). The existed parameters selection algorithms, such as the analytical, algebraic techniques and particle swarm optimization (PSO) algorithm, have their own shortcomings. In this paper, the problem of model selection for LS-SVM is discussed and a new method selecting the LS-SVM hyper-parameters is proposed based on the principles of the quantum-behaved particle swarm optimization (QPSO). The feasibility of this method is evaluated on data sets produced by sinc function. Experimental results show that LS-SVM of QPSO-based hyperparameters selection obtains better generalization capability and has more fast convergence speed than PSO-based hyper-parameters selection. Furthermore, the proposed method was applied to establish a softsensor model for content of Bisphenol A (CBPA) in rearrangement productive process. The results of real data simulation also show that this method is effective.

show abstract

“…Several approaches such as a manual method, a grid search method, a genetic algorithm, particle swarm optimization and their combinations with grid search [23,6,13,12,25,24,14,45] have been proposed for the selection of user-defined parameters with SVM. In the present study, a manual method (carrying out a large number of trials by using different combinations of user-defined parameters with both modelling approaches) was used to select user-defined parameters (i.e., C, c, r, x, e and Gaussian noise).…”

Section: Details Of Gp and Svmmentioning

confidence: 99%