Bayesian Penalized Method for Streaming Feature Selection

The online feature selection with streaming features has become more and more important in recent years. In contrast to standard feature selection method, streaming feature selection method can select feature dynamically without exploring full feature space. In some applications, when the information of all features is unknown in advance, standard feature selection methods cannot perform well in this setting. Moreover, in ultrahigh dimensional data analysis, especially considering interaction effects between features, the streaming feature selection method enjoy computational feasibility. In this paper, we proposed a Bayesian penalized method for streaming feature selection problem. The proposed approach adopts Bayesian regularization into penalized model which can adaptively estimates regularization parameter based on the coefficients of current model. Comparing with many existing streaming feature selection methods, our method can work for more general case of predictive model. The proposed method is evaluated extensively on various high-dimensional datasets. The experimental results show that the algorithm is competitive with many existing streaming and traditional feature selection algorithms.