Support vector machine for classification of text and call induced attention shift in driving using wireless EEG and imagined speech recognition concept

• Main Authors: Renosa, Claire Receli Morales, Renosa, Claire Receli M.
• Format: text
• Language: English
• Published: Animo Repository 2022
• Online Access: https://animorepository.dlsu.edu.ph/etdm_mem/6; https://animorepository.dlsu.edu.ph/cgi/viewcontent.cgi?article=1005&context=etdm_mem
• Institution: De La Salle University

For the past decade, inattentive or distracted driving has been the leading cause of most car accidents. This continuously grows into a larger threat to society’s road safety as we make the shift towards the so-called digital revolution. This is due to the fact that over the years, wireless devices like mobile phones managed to keep its spot in various statistics as one of the top reasons for causing distraction while driving. This has lead groups of developers and researchers to grow a significant amount of interest in creating solutions and precautionary systems for this specific problem. This paper revolves around the same issue and aims to develop a machine learning model that could detect the mental state of a driver, pertaining specifically to the shift of a person’s attention-focus by collecting data from a wireless EEG device and processing it using the imagined-speech concept. Compared to current methods of detecting attention-focus in existing driving monitoring systems today, this detection paradigm aims to utilize this specific approach and makes use of that as the key factor for the development of the machine learning model. With the insertion of the imagined-speech concept, the process of the model development begins by putting volunteer subjects in a stimulated driving simulation within a rudimentary simulator set-up. During these simulations, they will be asked to wear the Emotiv EPOC headset, a commercialized non-invasive wireless EEG device used to collect raw electroencephalography (EEG) signals. These signals will later be analyzed and processed to extract the common signal feature across all collected data and be used as the determinant for classification of mobile phone induced driving distraction focusing on texting, calling and phone alerts. The main purpose of the study is to create and find the best combination of datasets and preprocessing methods to reach high accuracy in stimuli classification. In conclusion, this research was able to achieve a good range of accuracy on varying factors considered in this study.