GA-SVM applied to the fall detection system
Hsi-Chun Kao, Jui‐Chung Hung, Chih-Pang Huang · 2017
This paper presents the new method use the smart watch to detect the fall system. Uses the Genetic Algorithm for Support Vector Machine (GA-SVM) algorithm to detect to fall or Activity Daily Living(ADL). We collect the data and transfer to the spectrum observations, we found that the attributes of the spectrum are the complex issue. We use the GA to do filter the useful features effectively and then SVM will help to set the classification for these features. The result shows that this method can avoid the Overfitting problem caused by a lot features and reduce the complexity of the classification. This method has higher accuracy than the traditional classifier such as SVM and C4.5. In addition, we apply this method to apply in the smart wristwatch application, it can auto send the information to the related friends when the fall event happened.