The hierarchical classification model using Support Vector Machine with multiple kernels in human behavioral pattern recognition
Sorin Soviany, Virginia Cristiana Sandulescu, Sorin Puşcoci · 2017
The paper proposes a classification model for human behavioral patterns recognition in which the decisions are provided based on several Support Vector Machines classifiers within a multi-level decision structure. SVMs are suitable for applications in which the input data feature spaces are very large, involving many features. The human behavior recognition is a relevant example of such application. On the other hand, the model proposes several kernels to be applied for SVMs in order to achieve improved performances for the real applications in which the human behavioral patterns recognition is required, as in the case of tele-assistance services.