Affect Classification with Physiological Significant Parameters and Support Vector Machine
Affect identification improves communication between humans and machines. Thus Human-Computer Interaction, which employed for many applications, will be more efficient if emotions incorporated in the design. Affect is an intelligent agent, and it is a complicated process to determine and categorize the emotional states. Emotional changes in human beings have a high impact on human health; therefore,categorizing emotions with high accuracy is essential. In this work, the principalaim is to enhance the emotional state classification efficiency.In this work, emotional states grouped into two as positive and negative. The output efficiently classifiedby having input data, which is accurate, features are appropriate, and the classification algorithm is efficient. The Electrocardiogram (ECG)dataset used in this work is from a standard physionetdatabase,and for classification,a dynamicclassifier Support Vector Machine used. The classification efficiency enhancedwith appropriate features of ECG physiological signal, which significantly contributestogrouping. Thus from this analysis, it can be concluded that twotime-domain characteristicsand three frequency-domain characteristics are dominant. Therefore a total of fiveelements can efficiently classify the emotional state into two groups as positive and negative.