A GA-based feature selection and parameter optimization for support tucker machine

Compare with vector, tensor can reserve structural information and tensor model algorithm can exploit such information. Unfortunately, tensor contain many redundant information which is undesirable to Support Tucker Machines(STuMs), therefore we present a genetic algorithm (GA) based algorithm to feature selection and parameter optimization simultaneously for the STuMs. The proposed algorithm can sweep away the irrelevant information in tensor data and obtain a better generalized accuracy. Experiments conducted on third-order gait recognition datasets to examine the performance of the proposed algorithm. The results show that proposed algorithm can provide a significant performance gain in terms of generalized accuracy and training speed for tensor classification.