Detection of autism through the machine learning and deep learning analysis of Magnetoencephalography scans.

Work done for pf. Karim Jerbi of the

Systems Neuroscience and Cognitive Neuroimaging Psychology Department University of Montreal http://www.karimjerbi.com/

This only includes the code. The data isn't currently directly available to the public.

Usage:

The brackets mean "pick one".

• python main.py single_dim_classification [SVM, KNN, RandomForests, SKL_LR]

• python main.py spatial [keras_dense, keras_conv, tflearn_resnet, tflearn_bn_vgg, tflearn_vgg, vgg, cnn, resnet]

• [Todo, Not fully functional] python main.py sequence_classification

There is a wide range of command line parameters to fiddle with everything. A few more details can be had by runnin python main.py --help.

Dependencies:

A whole bunch of stuff, which includes, non-exhaustively:

• numpy
• Tensorflow (tensorflow-gpu ideally)
• Tflearn
• Keras
• Scikit-learn
• mne

( Keywords : mne, magnetoencephalography, meg, python, machine learning, ml, deep learning, dl, scikit-learn, skl, support vector machines, svm, random forests, neural networks, nn, convnets, cnn, tensorflow, tf, neuroscience, neuroimaging, neuro psychiatry, resnet, densenet )