Enhancing interpretability of automatically extracted machine learning features: application to a RBM-Random Forest system on brain lesion segmentation

Machine learning approaches can be broadly divided into two categories: those using hand-crafted features, and those relying on Representation Learning techniques. Representation Learning refers to a set of general machine learning methods for automatic learning and extraction of features directly from data. By contrast, hand-crafted features require expert knowledge on the problem, hence making them more problem-dependent (LeCun et  al., 2015). Notwithstanding, there is usually a data representation mapping stage that takes the input data and transforms it into a more discriminative representation.

Source: Medical Image Analysis - December 19, 2017 Category: Radiology Authors: S érgio Pereira, Raphael Meier, Richard McKinley, Roland Wiest, Victor Alves, Carlos A. Silva, Mauricio Reyes Source Type: research