Feature selection and multivariate Gaussian probability distribution for user behavior recognition

Abstract

User behavior recognition using sensory data has become an active field of research in the domain of pervasive and mobile computing. The Principal Component Analysis (PCA) is a common method for feature selection. To obtain the best description and the best classification characteristics of different behaviors, an algorithm of Principal Component Analysis based on Regularized Mutual Information (RMIPCA) is presented. The new algorithm introduces the category information, and uses the sum of regularized mutual information matrices between features under different behavior to replace the covariance matrix. Furthermore, the extracted feature is calculated based on multivariate Gaussian probability distribution, and the transitional noise behavior data is removed according to the probability value. The simulation results show that the performance of the proposed is better than the others compared, which can effectively identify the user's daily behavior.