Deep CNN for classifying EEG Emotion

Performance

90.1% validation accuracy

you'll need the SEED I dataset. Application here: application form

Pytorch model design

• 5 blocks of 2 conv layer with LeakyReLU, maxpool, dropout, and residual connections
• LSTM with residual
• classification layers with dropout, ReLU
• softmax, ce

Use

• extract SEED I data to ./data
• you will want to alter the batch size and workers on the loader to suit your env
• ./CNN-LSTM-RES.sh