Your Mind is a Rat
or, using a neural network to classify EEG signals in
a maze game BCI
Electricity
Fast fourier transform decomposes raw signal
--FFT-->
Data samples
Baseline
7044:9689:11807:13458:13174
113325:102718:2949:29319:14806
Task
80259:521265:30119:9249:1319
29409:1510221:12880:2818:490
Neural network
1.Sampling
Lightweight iOS app to take 20-second samples, at 1 sample/sec, of baseline and
task brain actions
2. Training
PyBrain (Python-Based Reinforcement Learning, Artificial Intelligence and Neural
Network Library) to create and train a backpropagation neural net for classification
3. BCI / Game
● Create a neural network using training weights with MLPNeuralNet (multilayer
perceptron neural network library for iOS and OSX)
● Classify a number of samples from incoming EEG signal and take aggregate
● Use result to make binary move choice, alternating between left/right and
up/down

Your Mind is a Rat