Comparative assessment of forest optimization with novel deep ensembling technique for human activity recognition based on data collected from smartphones

Abstract

Human Activity Recognition is a technique for classifying a person's activity using sensitive sensors that are influenced by movement. Improving the performance of Human Activity Recognition based on information sensed by smartphones. In this we have considered two groups namely forest optimization with the sample size of 111 and novel deep ensemble technique with the sample size of 111. Accuracy is computed with the data set size of 428 to recognize the different human Activities (Hand Waving, running). It was observed that the Forest Optimization Algorithm obtains Accuracy of 95.75% and loss is 12.6%. Forest Optimization technique appears to have better significance than novel deep ensemble technique with value p=0.002. The Result proves the forest optimization approaches with varying seed value have significant improvement in human activity recognition.