Decode Brain Signal Into Thai Word Using EEG and L-SVM

This research decodes the brain signal into words with the help of machine learning and modern devices. It’s been a long time since researchers tried to understand brain signals. Scientists succeed in decoding the brain signal using invasive methods such as ECoG. However, this process requires a dedicated skull surgery operation which poses a high risk to the subject. To ensure general-purpose availability, we propose a healthy non-invasive Brain-Computer Interface (BCI) system to convert brain signals to words. In this paper, we show the experiment result that proved EEG equipment has enough quality to decode brain signals to text with high accuracy. Our method uses EEG raw offline data, and processes with Statistic Features and Linear Support Vector Machine (L-SVM) machine learning techniques from 2 Thai native persons with 6 Thai words including ถูกต้อง (Correct), ช่วยด้วย (Help me), หิวข้าว (Hungry), ไม่ใช่ (No), ปวดท้อง (Stomachache), หิวน้ำ (Thirsty) and we can achieve 100% accuracy.