Abstract
Falls in elderly people are a significant cause for injury. Effective prevention strategies are therefore helpful in addressing this problem. A number of machine learning approaches have been proposed for identification of near fall situations, thus by preventing fall related injuries. However, many of the existing algorithms are supervised and require long training time, especially on large datasets. This paper investigates training data subset selection using a well-known unsupervised algorithm K-means clustering. The effect of cascading the priori information obtained from K-means is evaluated using three supervised algorithms namely K-nearest neighbor, Decision-Tree and Random forest. Experimental results illustrate that computational time is reduced significantly and the fall recognition rate is preserved on the reduced training dataset.
Recommended Citation
Sakurai, Dr. Kouichi; Mandal, Dr. Lopa; Das, Dr. Baisakhi; and Jothi, R.
(2024)
"An empirical study on fall-detection using K-means based training data subset selection,"
American Journal of Advanced Computing (AJAC): Vol. 1:
Iss.
1, Article 2.
Available at:
https://research.smartsociety.org/ajac/vol1/iss1/2