As a guest user you are not logged in or recognized by your IP address. You have
access to the Front Matter, Abstracts, Author Index, Subject Index and the full
text of Open Access publications.
This paper considers the feature selection scenario where features are generated sequentially and visible one by one and the full feature space is unknown in advance. Although there are some existing online streaming feature selection algorithms based on statistics or optimization, the main disadvantage they suffer is that a feature is permanently discarded once it is considered irrelevant even though it may become more relevant owing to the changing feature space. To overcome this drawback, our paper proposes a new online streaming feature selection algorithm based on a fixed-size buffer pool (BFS). Specifically, BFS maintains a buffer pool to dynamically retain and retrieve features to deal with the changing feature space and combines two different feature selector by a boost manner to improve the predictive performance. Extensive experiments are conducted on real-world datasets to evaluate the effectiveness of BFS and the results suggest that BFS is comparable to the state-of-the-art streaming feature selection algorithms, and requires less features or achieves higher predictive accuracy.
This website uses cookies
We use cookies to provide you with the best possible experience. They also allow us to analyze user behavior in order to constantly improve the website for you. Info about the privacy policy of IOS Press.
This website uses cookies
We use cookies to provide you with the best possible experience. They also allow us to analyze user behavior in order to constantly improve the website for you. Info about the privacy policy of IOS Press.