Reproduce “MontageFlow: Physically Valid High Fidelity EEG Generation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Sudarshan Regmi, Nabaraj Subedi, Arvind Pillai, Yu Yvonne Wu, Yuliang Chen, Michael V. Heinz, Lisa Marsch, Nicholas C. Jacobson and 1 more
MontageFlow generates 16-channel EEG by projecting starting noise and each predicted change into the 14-dimensional subspace satisfying two montage loop constraints, with a window-level vector to preserve variation. On TUAB, its test TS-FID is 59.9 versus 149.1 for vanilla diffusion, with montage indices of 0.000 and 0.071; MontageDiff scores lower on TS-FID. Results cover three Temple University Hospital corpora.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.