A Novel Two-Level One-vs-Rest Classifier
Tao Wang, Jiang Yongjia, Rong Xiangsheng · 2019
Classification plays a significant role in pharmacokinetics, pharmacology dynamics and etc. This paper proposes a novel Two-Level assembling Classifier (TLC) that fulfills label detection with two steps, with aim to achieve higher accuracy and feasibility. TLC constructs the micro classifier based on local boundary information, combines them in an informed way to form basic classifiers and integrates the local decisions of basic classifiers to give the global decision. TLC is equipped with parameterization heuristics to foster computation. Empirical evidence on real datasets demonstrates the performance of TLC compared with the peers.