Is end-to-end learning possible? Is it possible to classify MI-eeg by end-to-end learning without extracting features from EEG signals obtained from opebci cython?
Of course, deep learning is good at learning arbitrary relationships between input and output data. The problem becomes
1) Do you have enough data (often you can over fit to single trials)?
2) Do you understand the data and how to apply appropriate models?
3) Do you have access to enough compute to run and train the model?
4) I'm sure there is something I'll forget, consider this one everything you run into I don't mention or get wrong lol
A key thing to remember here is that instead of providing filtering and features from hand crafted algorithms the model must now learn to pick out the important information itself and you will need to scale and chose the network architecture accordingly.
Comments
Of course, deep learning is good at learning arbitrary relationships between input and output data. The problem becomes
1) Do you have enough data (often you can over fit to single trials)?
2) Do you understand the data and how to apply appropriate models?
3) Do you have access to enough compute to run and train the model?
4) I'm sure there is something I'll forget, consider this one everything you run into I don't mention or get wrong lol
A key thing to remember here is that instead of providing filtering and features from hand crafted algorithms the model must now learn to pick out the important information itself and you will need to scale and chose the network architecture accordingly.
Good luck and let me know how it goes!