Enhancement of Accuracy in Speech Recognition System for Detection of Covid-19 using Support Vector Machine
Rallapalli Jhansi, G. Uganya
Abstract
The goal of this study is todetect Covid-19 by innovative speech recognition using Support Vector Machine (SVM) and comparing its accuracy with Random Forest (RF). Speech recognition using SVM is considered as group 1 and Random Forest is considered as group 2, where each group has 20 samples. In order to solve classification and regression problems, a machine learning technique known as RF is employed. Using the average of the samples for regression and the majority vote for classification, it generates decision trees. A machine learning algorithm called SVM used to identify the classification of patients based on genes and other biological problems. These groups were analysed by an independent sample T -test with alpha is 0.05; G power is 80% and confidence interval is 95%. RF achieves an accuracy of 84.5% and SVM achieves an accuracy of 94.6% with significance value 0.045 (p<0.05). From statistical analysis, it is observed that SVM has significantly better accuracy than RF.