No abstract

No abstract

No abstract

This research proposed an efficient method in classification of Human Activity Recognition tasks. The evaluated tuned models show higher than 99 percent mean accuracy and gain more training and testing accuracy in comparison to previous studies. Dimensionally reduction have been introduced based on P-value evaluation in feature space. Finally a hybrid model that compressed statistically in feature space alongside with Neural Network architecture have been proposed. The final model could be used as best architecture of hardware implementation in gesture recognition applications.

No abstract