P300 ERP Component on Eating Habits Profiling Using Dynamic Evolving Spiking Neural Network (deSNN)

Abstract

— Unhealthy eating habits have become a big issue that often causes many chronic diseases in various countries in recent years. The current assessment to identify the status of eating habits is to use self-assessment. However, self-assessment is known to have an error or uncertainty value due to cognitive factors from respondents that affect the results of the assessment. A person's profile is potentially measured by reviewing Event-related potential (ERP) which is an ideal technique for understanding perception and attention. This study uses images of healthy and unhealthy foods as a stimulus when recording EEG data. The method used for classification is dynamic evolving spiking neural network (deSSN) based on the Neucube architecture. The results showed that the mean amplitude of the P300 component discovered in the Parietal and Occipital lobes was higher for healthy food in the healthy eating habits group. Whereas the unhealthy eating habits group was higher for unhealthy foods. The deSNN classification is proven to operate in learning ERP data but the accuracy rate is not too high due to inadequate sample training