Wearable Sensor-Based Human Activity Recognition for Smart Healthcare and Behavior Monitoring over Efficient Feature Selection and MLP

Human Activity Recognition (HAR) is the fundamental area of artificial intelligence, machine learning, and deep learning for all classification approaches dealing with human actions.The current HAR system encompasses advanced information preprocessing alongside customized classification features and algorithms.Widespread use of 4th-order median filtering is a primary noise reduction technique before signal enhancement through Hamming window processing.The source data uses Particle Swarm Optimization (PSO) to determine optimized characteristics that form the basis of retention discrimination from other features.The system employs Multi-Layer Perceptron (MLP) technology, which supports deep learning framework-driven activity classification operations.The system's effectiveness was evaluated on three prominent datasets: HCI, HMP, and WISDM.When tested on HCI data, the proposed approach achieved 85% precision rates but recorded 94% accuracy for HMP data preceding WISDM-based recognition at 92%.These results announce the capacity of the system for truthful and trustworthy activity cataloging in varied real-world artificial intelligence products.