Tailored and transfer learning

Module API changes

• Create _train_encdec private function to train a self-supervised encoder-decoder neural network for regression
• Create _test_encdec private function to test a self-supervised encoder-decoder neural network
• Create _proxy_optim private function to train and optimise a self-supervised encoder-decoder neural network
• Create deep_transfer_learn function to perform transfer learning (self-supervised pretext task and transfer to downstream classification task)
• Add sort parameter to process_epochs for sorting channels by chromophore type (HbO, HbR)

Main script changes

• Change path of extra_stats.py
• Create comparison_stats_dataset.py to compare models on a dataset with statistical tests
• Create comparison_stats_task.py to compare models on multiple datasets with statistical tests
• Create tailored_generalised.py (subject-independent n-back classification)
• Create tailored_window_size.py (subject-independent n-back classification with different window sizes)
• Create tailored_shin_nb.py (subject-independent classification on Shin et al., 2018 n-back task)
• Create transfer.py (transfer learning with labelled and unlabelled data)
• Create transfer_no_unlab.py (transfer learning control)

Other changes

• Checklist updated as a list of questions
• Version number