The Use of Kernel-Based Extreme Learning Machine and Well-Known Classification Algorithms for Fall Detection

Abstract

Fall is a common occurrence in older people and causes dangerous consequences. Therefore, the fall detection needs to be done quickly and efficiently. In our developed expert system, three different algorithms have been tried to find fall detection by classification method. Two of these algorithms are the widely used naive Bayesian and k-nearest neighbor algorithms for fall detection. The other used algorithm is kernel-based extreme learning machine algorithm. The main purpose of this study is to compare these three algorithms in detecting falls. As a result of the work, the kernel-based extreme learning machine algorithm has been shown to find the fall detection with better results than the k-nearest neighbor and naive Bayesian algorithms. The kernel-based extreme learning machine algorithm achieves a better F-measure value of 3.5% according to the k-nearest neighbor algorithm (k = 1), 1.8% according to the k-nearest neighbor algorithm (k = 5) and 6.6% according to the naive Bayesian algorithm.