Innovative Approaches to Gender Classification through Unsupervised Machine Learning and Multi-Activity Fusion

In the last decade, gender recognition has garnered significant attention for its diverse applications in healthcare, sports, rehabilitation, and wearable electronics. This study offers a wearable sensor device to record various activities using inertial measurement units as part of its gender classification process. During seven activities that entail standing, walking, and climbing, the system uses five sensors on both the upper and lower bodies. The gender recognition model is developed through a thorough investigation, utilizing unsupervised machine learning methods. Specific sensor placements and behavioral patterns are identified to enhance accurate gender classification using unsupervised machine learning. Applying Kmeans Clustering and Gaussian Mixture Models (GMM) for gender classification in a single activity achieves a maximum accuracy of 86.42%. Multi-activity gender classification surpasses this, demonstrating a notable accuracy of 90.14%, particularly in the combination of activities, Romberg test eyes open, Single leg stance eyes closed, Walking, and Staircase up and down.