Feature Selection and Analysis on Correlated Breath Data
Citations
6 citations
Additional excerpts
...The stability and effectiveness of the SVM-RFE+CBR ensemble method have already been verified [38]....
[...]
...To remove the substantial irrelevant connectivity and avoid the possible overfitting issue due to the fact that the number of features is much larger than that of samples, linear support vector machine recursive feature elimination (SVM-RFE) with correlation bias reduction (CBR) [37] was utilized....
[...]
...Given that correlations of FC in brain network may cause the importance of features to be underestimated, the CBR method was employed to reduce this correlation bias....
[...]
...Subsequently, the SVM-RFE+CBR method was applied in all the data, and two ranked feature sets of each session were obtained based on the significance of each feature....
[...]
References
15,696 citations
14,509 citations
8,078 citations
7,939 citations
7,075 citations