PyMVPD: MultiVariate Pattern Dependence in Python

1. Linear Regression Models

• L2_LR: linear regression model with L2 regularization
• PCA_LR: linear regression model with no regularization after principal component analysis (PCA)

2. Neural Network Models

• NN_1layer: 1-layer fully-connected linear neural network model
• NN_5layer: 5-layer fully-connected linear neural network model
• NN_5layer_dense: 5-layer fully-connected linear neural network model with dense connections

Data of one subject from the StudyForrest dataset: FFA - fusiform face area, GM - grey matter.

• Raw data were first preprocessed using fMRIPrep and then denoised by using CompCor (see more details in Fang et al. 2019).

1. Choose one MVPD model, set model parameters, input functional data and ROI masks, set output directory in analysis_spec.py;
2. Run data_loading.py to preprocess functional data;

python3 data_loading.py

3. Run MVPD model:

sh analysis_exec.sh

Reach out to [email protected] for questions, suggestions and feedback.