neuroIN is a deep learning package for neuroimaging data.
The basic workflow for neuroIN is: -process EEG (other modalities to come) data into image and video like arrays -initialize an architecture for your data -train network and optimize hyperparameters -visualize results and network features
neuroin.io contains functions and classes associated with importing datasets from neuroimaging data formats into NumPy arrays. Information on imported datasets is stored in configuration files that are used to load Dataset objects so imported datasets can be used.
neuroin.preprocess contains functions for preprocessing imported data; this entails training/testing split generation, data normalization, and data augmentation methods.
neuroin.models contains classes for the different architectures supported by the package.
neuroin.training contains function to train models, classify data, log training history, and save trained models.
neuroin.optim contains functions for optimizing network hyperparameters.
neuroin.vis contains visualization functions for training results and feature visualization.