A Novel Martingale Model for Human Activity Recognition using Robust Optimisation Techniques

The subject of human activity recognition is becoming eminent in the study of human movement patterns and disabilities. Many existing approaches are effective in identifying change points in physical activities; however, algorithm optimisation could potentially be applied to improve some of these models. In this paper, we proposed a method known as mean of the geometric moving average of the Martingale sequence that can detect changes in human activity recognition. Furthermore, the suggested method is optimised for enhanced performance using meta-heuristic optimisation techniques based on the genetic algorithm (GA) and particle swarm optimisation (PSO) algorithms. Experimentation shows that the suggested method improves the performance of the previous Martingale method.