This is a great task for someone with no GPU access.
There are a number of open source fMRI datasets available that either have trials corresponding to "simple" stimuli that we want to be able to decode, or have imagery trials we can use to train on. In order to utilize this data, we need to be able to easily load this data in a preprocessed (Z-scored) fashion. See the dataset code I wrote for the UMN datasets in the dataset_preparation branch, and the data loading methods in utils for inspiration.
This is a great task for someone with no GPU access.
There are a number of open source fMRI datasets available that either have trials corresponding to "simple" stimuli that we want to be able to decode, or have imagery trials we can use to train on. In order to utilize this data, we need to be able to easily load this data in a preprocessed (Z-scored) fashion. See the dataset code I wrote for the UMN datasets in the dataset_preparation branch, and the data loading methods in utils for inspiration.
Some candidates for datasets, the list is not exhaustive, feel free to search for others. Generic Object Decoding Deep Image Reconstruction NeuroMod CMI Neuroimaging Dataset