Incremental Framework for Feature Selection and Bayesian Classification for Multivariate Normal Distribution
Ramesh Kumar Agrawal, Manju Bala, Rajni Bala · 2009
In this paper, an incremental framework for feature selection and Bayesian classification for multivariate normal distribution is proposed. Feature set can be determined incrementally using Kullback divergence and Chernoff distance measures which are commonly used for feature selection. The proposed integrated incremental learning is computationally efficient over its batch mode in terms of time. The effectiveness of the proposed method has been demonstrated through experiments on different datasets. It is found on the basis of experiments that the new scheme has an equivalent power compared to its batch mode in terms of classification accuracy. However, the proposed integrated incremental learning has very high speed efficiency in comparison to integrated batch learning.