The main objective of this project is to use supervised learning methods in machine learning to decode the class of images shown to subjects based on the subjects’ brain activity. Ideally, it would be possible to detect which features of brain activity contain most information about the visual stimulus presented to the subject.

The data set provided consists of the neural responses of showing 269 images to 74 distinct subjects. For each subject, there are multiple tests performed, using 33 distinct Brodmann areas for the electrodes; there are 554 subject and electrode area pairs provided in total. The images are classified into 6 distinct classes. Each neural response is denoted as a single number (power at the gamma band for a certain time window after the image is presented).

Python 2.7 and the scikit-learn package are used for the project.

The team working on the project consists of four members (in alphabetical order):

• Markus Hofmann (Computer Science, 2nd year Master’s exchange student)
• Gerrerth Kaur (Computer Science, 1st year Master’s student)
• Kevin Kremer (Computer Science, 2nd year Master’s exchange student)
• Ants-Oskar Mäesalu (Computer Science, 2nd year Master’s student)