Resources for the paper titled "EEG-GCNN: Augmenting Electroencephalogram-based Neurological Disease Diagnosis using a Domain-guided Graph Convolutional Neural Network". Accepted for publication (with an oral spotlight!) at ML4H Workshop, NeurIPS 2020.
Authors: Neeraj Wagh, Yogatheesan Varatharajah
Affiliation: Department of Bioengineering, University of Illinois at Urbana-Champaign
$ python heldout_test_run.py
to run the saved 10 final models on the held-out 30% test set of subjects using the pre-computed features. For trivial classifiers, run $ python chance_level_classification.py
. The mean and standard deviation values reported in Table 2 of the paper will be printed at the end of execution. See notes below for more details.After removing type cast error in edge weights calculation, the performance stays similar to what is published in the paper.
AUC | Precision | Recall | F-1 | Bal. Accuracy | |
---|---|---|---|---|---|
Shallow EEG-GCNN | 0.871(0.001) | 0.989(0.003) | 0.677(0.018) | 0.804(0.011) | 0.810(0.003) |
Deep EEG-GCNN | 0.863(0.001) | 0.988(0.006) | 0.660(0.031) | 0.791(0.020) | 0.800(0.002) |
Wagh, N. & Varatharajah, Y.. (2020). EEG-GCNN: Augmenting Electroencephalogram-based Neurological Disease Diagnosis using a Domain-guided Graph Convolutional Neural Network. Proceedings of the Machine Learning for Health NeurIPS Workshop, in PMLR 136:367-378 Available from http://proceedings.mlr.press/v136/wagh20a.html.