Multi-view support vector machines with sub-view learning

Multi-view learning improves the performance of existing tasks by using complementary information between multiple feature sets. In latest research, multi-view model privileged is proposed; specific models are PSVM-2V and MCPK. these models, views complement each other acting as policies; however, a single view contains that can guide classifier, framework does not consider it. order to use this correct support vector machine we propose for generating series small-scale based on hidden in view, which extends original parallel structure hierarchical with sub-view mechanism. paper, two structures SL-PSVM-2V SL-MCPK constructed. The new fully exploit data features view. Similarly, they follow principles consistency complementarity. We standard quadratic programming solver solve model. 55 groups classification experiments, proposed accuracy about 1.91% basis. ranks 1.3846 average experiment, indicating have good ability. addition, computational time statistics noise set experiments carried out prove effectiveness method from perspectives.