Implementation for the paper https://dl.acm.org/citation.cfm?doid=3240508.3240641

• Download the EEG data from here
• Extract it and place it in the project folder (.../ThoughtViz/data)

• Downwload the images used to train the models from here
• Extract them and palce it in the images folder (.../ThoughtViz/training/images)

• Download the trained EEG Classification models from here
• Extract them and place in the models folder (.../ThoughtViz/models/eeg_models)
• Download the trained image classifier models used in training from here
• Extract them and place in the training folder (.../ThoughtViz/training/trained_classifier_models)

1. EEG Classification

2. GAN Training

• Download the sample trained GAN models from here

• Extract them and place in the models folder (.../ThoughtViz/models/gan_models)

• Run test.py to run the sample tests

  1. Baseline Evaluation

     • DeLiGAN : Uses 1-hot class label as conditioning with MoGLayer at the input.

  2. Final Evaluation

     • Our Approach : Uses EEG encoding from the trained EEG classifier as conditioning. The encoding is used as weights in the MoGLayer

NOTE : Currently we have uploaded only one baseline model and our final model. Other models can be obtained by running the training code.