Optimal feature selection using PSO with SVM for epileptic EEG classification

Murugavel, A. S. Muthanantha; Ramakrishnan, S.

doi:10.1504/ijbet.2014.066229

Cited by 5 publications

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The experiments show these learning parameters in SVM, including boundary of Lagrange multiplier C, the condition parameter of convex quadratic optimization λ and ε-neighborhood parameter around solutions ε, have no obviously effect on the output results, but the kernel function parameter σ that have the largest influence on the output results is often difficult to identify only by trial and error [20]. Taking MSETD as fitness function, the kernel function parameter σ is optimized by virtue of PSO global search performance for optimal solutions and then the proper offset b and weight coefficient  are found, so that output results are optimal or suboptimal to meet the precision and accuracy of system measurement.…”

Section: The Pso Optimization Algorithm Designmentioning

confidence: 99%

Study on SVM Temperature Compensation of Liquid Ammonia Volumetric Flowmeter Based on Variable Weight Pso

Lin¹,

Ping²,

Gao³

et al. 2015

IJHT

View full text Add to dashboard Cite

When the quality of liquid ammonia is measured by volumetric flowmeter, the traditional quadratic expression method can't meet the accuracy of temperature compensation in modern coal chemical industry. So the temperature compensation method by support vector machine (SVM) regression is presented, and kernel function parameter σ of SVM is optimized by variable weight particle swarm optimization (PSO). After the performance analysis and comparison in PSO, the suitable linear inertia weight method is selected. Experimental results show that the temperature compensation accuracy of the SVM method based on the variable weight PSO is significantly higher than that of the traditional quadratic expression method.

show abstract

Section: The Pso Optimization Algorithm Designmentioning

confidence: 99%