Heterogeneous Ensemble Classification for Pre-miRNA with Discriminative Features
A heterogeneous ensemble algorithm which is the voting of multi-expert classifiers, SVM, kNN, and RF, in the task of pre-miRNA classification. Our model is trained on both human and plant pre-miRNA data and utilize robustness derivative features which has shown to be well distinquishing both plant and animal pre-miRNA hairpins from other non pre-miRNA hairpin-forming sequences.
Download: Sourcecode , readme.txt
Heterogeneous ensemble approach with discriminate features and modified-SMOTEbagging. Nucleic Acids Research, 2012, 1-12 [doi:10.1093/nar/gks878]