I’m Cherukuri Gowtham , a computer science student at VIT University, excited to apply for GSoC 2025 with MNE-Python under PSF. I’ve been coding in Python and have experience building an ML project I’m passionate about applying ML to real-world problems, and EEG analysis caught my interest.
For GSoC, I’d like to propose enhancing MNE’s interoperability with Braindecode by integrating an EEG classifier (e.g., using scikit-learn’s SVM) to decode signals like motor imagery or event responses. This builds on your “Improve interoperability with braindecode” idea, aiming to streamline preprocessing-to-decoding workflows for researchers. I’ve started exploring MNE’s sample data and Braindecode’s docs to get familiar.
I’d love to hear your thoughts! Are there mentors (e.g., Eric Larson, Dan McCloy) available to discuss this? Any suggestions for a good first contribution to get started? Thanks for your time—I’m thrilled to connect with this community!
Hi @cherukurigowtham, thanks for your interest in MNE-Python! The proposal document didn’t allow comments, so here are a few:
Integrate an SVM classifier (scikit-learn) for EEG decoding
MNE-Python already has the mne.decoding submodule, can you elaborate on how what you propose extends and fits in with what we already have?
refactor MNE-BIDS dataset handling for consistency
You’ll need to say more here. The BIDS standard is pretty rigorous/rigid, and a separate package (MNE-BIDS) already exists to wrangle BIDS-formatted datasets. What exactly is inconsistent, and what would you change?
Provide documentation and a demo notebook
MNE-BIDS-Pipeline is a command-line tool, it’s not really meant to operate in a Jupyter notebook.
Also please bear in mind the note at the top of our GSoC ideas page:
If you are not currently pursuing research activities in MEG or EEG and do not use or do not plan to use MNE-Python for your own research, our GSoC might not be for you. Our projects require domain-specific interest and are not simple coding jobs.
Are you already working with EEG / MEG data for your own research / studies? Are you using MNE-Python and related tools to do that work? If so, tell us about it in the proposal.
1.I’ve integrated mne.decoding.CSP with Braindecode’s ShallowFBCSPNet
2.specified refactoring MNE-BIDS for BaseDataset compatibility,
3.switched to a script demo.
I’m new to EEG but plan to use MNE for motor imagery research post-GSoC—details added. Does this address your points? Also working on a doc PR this week. Thanks!
new updated proposal: [https://docs.google.com/document/d/1NZ_4LiezoOt0BQx_iFnBFbZ-Yaj-bAKRAzSxIrYvMmE/edit?usp=sharing]