Spiking Neural Networks Trained with Particle Swarm Optimization for Motor Imagery Classification

Spiking neural networks (SNN) have been successfully applied in pattern classification problems. However, their performance for solving complex problems such as electroencephalography (EEG) classification has not been widely assessed. It is necessary to consider new approaches to select relevant information and for training SNN in order to improve their accuracy when applied to complex data classification. In this paper, we present a novel channel selection and classification method based on SNN trained with Particle Swarm Optimization (PSO) for the classification of EEG signals associated to motor imagery. The proposed method was able to correctly identify the most relevant channels for different motor imagery tasks. The SNN trained with PSO achieved good classification performances for a well-studied public database using a minimal number of EEG channels, showing advantages against other approaches, regarding both performance and system requirements.

[1]  J. Pernier,et al.  Oscillatory γ-Band (30–70 Hz) Activity Induced by a Visual Search Task in Humans , 1997, The Journal of Neuroscience.

[2]  Gérard Dreyfus,et al.  Neural networks - methodology and applications , 2005 .

[3]  Eugene M. Izhikevich,et al.  Simple model of spiking neurons , 2003, IEEE Trans. Neural Networks.

[4]  R. Ramírez-Mendoza,et al.  Motor imagery based brain–computer interfaces: An emerging technology to rehabilitate motor deficits , 2015, Neuropsychologia.

[5]  Thomas H. Wonnacott,et al.  Introductory Statistics , 2007, Technometrics.

[6]  Rui Zhang,et al.  Using particle swarm to select frequency band and time interval for feature extraction of EEG based BCI , 2014, Biomed. Signal Process. Control..

[7]  Beatriz A. Garro,et al.  Training Spiking Neural Models Using Artificial Bee Colony , 2015, Comput. Intell. Neurosci..

[8]  H. Sompolinsky,et al.  The tempotron: a neuron that learns spike timing–based decisions , 2006, Nature Neuroscience.

[9]  David M. W. Powers,et al.  PSO-based dimension reduction of EEG recordings: Implications for subject transfer in BCI , 2013, Neurocomputing.

[10]  Wulfram Gerstner,et al.  Spiking Neuron Models , 2002 .

[11]  Roberto Antonio Vázquez,et al.  Tuning the parameters of an integrate and fire neuron via a genetic algorithm for solving pattern recognition problems , 2015, Neurocomputing.

[12]  Kwee-Bo Sim,et al.  Analysis the effect of PCA for feature reduction in non-stationary EEG based motor imagery of BCI system , 2014 .

[13]  L. Abbott,et al.  Cortical Development and Remapping through Spike Timing-Dependent Plasticity , 2001, Neuron.

[14]  Yue Shi,et al.  A modified particle swarm optimizer , 1998, 1998 IEEE International Conference on Evolutionary Computation Proceedings. IEEE World Congress on Computational Intelligence (Cat. No.98TH8360).

[15]  Qingguo Wei,et al.  Binary particle swarm optimization for frequency band selection in motor imagery based brain-computer interfaces. , 2015, Bio-medical materials and engineering.

[16]  Sander M. Bohte,et al.  Error-backpropagation in temporally encoded networks of spiking neurons , 2000, Neurocomputing.

[17]  Subhojit Ghosh,et al.  Classification of Two Class Motor Imagery Tasks Using Hybrid GA-PSO Based K-Means Clustering , 2015, Comput. Intell. Neurosci..

[18]  Maurice Clerc,et al.  The particle swarm - explosion, stability, and convergence in a multidimensional complex space , 2002, IEEE Trans. Evol. Comput..

[19]  Beatriz A. Garro,et al.  Designing Artificial Neural Networks Using Particle Swarm Optimization Algorithms , 2015, Comput. Intell. Neurosci..

[20]  G. Pfurtscheller,et al.  The BCI competition III: validating alternative approaches to actual BCI problems , 2006, IEEE Transactions on Neural Systems and Rehabilitation Engineering.