Optimization with Label Weighting Extreme Learning Machine for Human Activity Recognition
L. Maria Anthony Kumar, S. Murugan · 2023
Human Activity Recognition (HAR), is a field of study related to the spontaneous detection of day-to-day activities performed by people based on time series data using sensors. Over the past few years, there have been numerous progressions in interconnected sensing technology namely edge computing, cloud, sensors, and IoT. In recent times, deep learning (DL) based method has gained popularity for HAR because they utilize representation learning technique that could identify hidden patterns in data and could automatically generate optimum features from raw input datasets produced from sensors without human interference. This study introduces a new Fire Hawks optimization with Label Weighting Extreme Learning Machine (FHO-LWELM) algorithm for HAR The presented FHO-LWELM technique aims to identify and classify different kinds of human activities. To do so, the presented FHO-LWELM technique exploits the LWELM model for recognition purposes. At the same time, the parameters involved in the LWELM model can be optimally chosen by the use of FHO algorithm and it led to improved recognition rate. The performance validation of the FHO-LWELM technique is examined on HAR dataset and the experimental outcomes stated the improved outcomes of the FHO-LWELM technique.