A hierarchical fusion framework integrating random projection-based classifiers: application in head and neck squamous carcinoma cancer

Ensemble methods achieves better performance than single classifier model. Classifier diversity and fusion architecture are equally important for building a successful multi-classifier system. In this study, we introduced random projection to obtain the required classifier diversity and then proposed a hierarchical framework, namely a novel hierarchical fusion integrating random projection diversified classifiers (HFRPC). The proposed hierarchical fusion scheme was validated on survival prediction of head and neck squamous carcinoma cancer (HNSCC). Experimental results have demonstrated the superiority of the proposed HFRPC framework over the base classifier member and the state-of-the-art benchmark ensemble methods, rendering it a potential tool to assist medical decision making in the practical clinical setting.