jetW/Brain-Computer-Interface

This project introduces a way to establish direct communication pathway between brain and computers. With labeled data, I implement a SVM classifier to classify the brain signals into two clusters when decoding the movement direction, left or right. The most work of this project is to implement a SVM classifier. First, this classification problem is converted into an unconstrained non-linear optimization problem. Interior point method is utilized to solve the logarithmic barrier. Inside the IPM algorithm, I use Newton method to solve the optimization problem in every loop. The stopping criterion of Newton method is by line search. Through this process, the optimal solution is rely on the value of  . Therefore, two fold cross validation is built inside the program to determine the optimal  .

MATLAB