Prediction of Human Activity Recognition Using Convolution Neural Network Algorithm in Comparison with Grid Search Algorithm

Abstract

The objective of this piece is to find human activity recognition not through the use of grid search but rather through the application of convolution neural network algorithms. The calculation is carried out by utilising G-power 0.8 with alpha, and the confidence interval is established at 95%. Fifty people will serve as the sample size for the algorithm that uses convolution neural networks to make predictions about human activity recognition (Group 1 equals twenty-five, and Group 2 equals twenty-five). In comparison, the accuracy that can be achieved through grid search is 89.6012, while the accuracy that can be achieved through the Novel Convolution Neural Network is 98.6512. The performance of the Novel Convolution Neural Network is noticeably superior to that of grid search because it incorporates the accuracy of both methods into a single solution.