Matrix-Autoencoder

Code for Matrix Autoencoder paper in https://arxiv.org/pdf/2105.14409.pdf

Code in Branch Master

INSTRUCTIONS:

1. Read Appropriate Dataset from Data: HCP/ASD -- modify line 55 in Main.py (ASD - num_classes =3, HCP - num_classes = 1 )
2. Change Parameters according to model (default setting) or baselines (i.e. set lambda_0/1/gamma = 0 for no fMRI/no DTI/Decoupled ANN*) + uncomment lines 95 -100 in Main.py for ANN baseline
3. Open Terminal Run Main.py

ORGANISATION:

~/Code

1. Main.py - runner script
2. Train_Module.py - scripts for training models
3. SPD_net_model.py - script for main model definition

~/Data:

1. /Data/ASD: Data_ASD.mat - ASD dataset

   /Outputs /SPDNet /Baselines /Train /Test

2. /Data/HCP Data_HCP.mat - dataset for training and testing Data_HCP_val.mat - dataset for validation

   /Outputs /SPDNet /Baselines /Train /Test /Validation