Tree based feature induction for biomedical data

During the recent years, a great advance in both biomedical data acquisition technologies and feature extraction methods has been witnessed. Harnessing these new tools and technologies has led to an indisputable increase in the number of available biomedical datasets. Despite the efforts made so far, the representational power of features used to describe a sample in such datasets, such as a gene in gene function prediction datasets or a protein in protein interaction datasets has yet to be improved. Here, the performed study focuses on the feature representation power from a machine learning perspective.status: publishe