Overhead Reduction in EEG signals using Particle Swarm Optimization and Independent Component Analysis
|Author(s)||Abdolreza Asadi Ghanbari, Karim Adinehvand, Mousa Mohammad Nia|
|Issue Date||May, 2014|
|Publishing Date||May, 2014|
|Keywords||particle swarm optimization (PSO); Independent component analysis, Artifact; Feature selection (FS).|
Feature selection (FS) is a global optimization problem in machine learning, which reduces the number of features, removes irrelevant, noisy and redundant data, and results in acceptable recognition accuracy. It is the most important step that affects the performance of a pattern recognition system. This paper presents a novel feature selection algorithm based on particle swarm optimization (PSO). PSO is a computational paradigm based on the idea of collaborative behavior inspired by the social behavior of bird flocking or fish schooling. The algorithm is applied to coefficients extracted by feature extraction technique: the discrete wavelet transform (DWT). The proposed PSO-based feature selection algorithm is utilized to search the feature space for the optimal feature subset where features are carefully selected according to a well defined discrimination criterion. Evolution is driven by a fitness function defined in terms of maximizing the class separation. The classifier performance and the length of selected feature vector are considered for performance evaluation.
Key performance characteristics of BCI systems are speed (i.e., how long it takes to make a selection) and precision (i.e., how often the executed selection is the one the user intended). Current systems allow for one selection within several seconds at a relatively high accuracy. Expressed in bit rate, which combines both speed and accuracy, the sustained performance of typical non-invasive and invasive BCI systems is still modest. Artifacts and Redundancies with acquired data are two major reasons for this limited capacity of Current BCIs. Artifacts are undesired signals that can introduce significant changes in brain signals and ultimately affect the neurological phenomenon. In new BCI systems for increase accuracy, increased number of electrodes. In this case the increased number of electrodes causes a non-linear increase Redundancy. This article used PSO for best feature selection and independent component analysis (ICA) for artifacts removal in EEG signal and Redundancy Reduction.
Experimental results show that the PSO-based feature selection algorithm was found to generate excellent classification results with the minimal set of selected features.