Noise benefits in motor imagery classification using ensemble support vector machine
Rujipan Sampanna, Sanya Mitaim · 2014
This paper explores how noise can improve classification accuracy of motor imagery classification using an ensemble support vector machine (ESVM) classifier. We add white Gaussian noise to the EEG signals and use them with the original signal data set for the ESVM training process. The ESVM classifier uses coefficients of the discrete wavelet transform (DWT) and coefficients of the autoregressive (AR) model as features for classification. Experimental results show that training ESVM with concatenated original data set and noise-added data sets can improve the classification accuracy. The classification system attains maximum accuracy when noise intensity is not zero and thus the system shows the stochastic resonance effect.