A new method for motor imagery classification based on Hidden Markov Model

Hidden Markov Model (HMM) has already been used to classify EEG signals in the field of Brain Computer Interfaces (BCIs). In many conventional methods, the Expectation-Maximization (EM) algorithm is used to estimate the HMM parameters for EEG classification. The EM algorithm is an iterative method for finding Maximum Likelihood (ML) or Maximum A Posteriori (MAP) estimates of parameters in statistical models. However, it can be easily trapped into a shallow local optimum. Recently, large margin HMMs is used to obtain the HMM parameters based on the principle of maximizing the minimum margin and it has been applied successfully in speech recognition. Inspired by it, we propose to use the large margin HMMs method in classification of EEG signals about motor imagery by establishing HMMs for different types of signals. Experimental results demonstrate that HMM parameters estimation via the new method can significantly improve the accuracy of motor imagery classification.