Skip to content

Latest commit

 

History

History
17 lines (12 loc) · 940 Bytes

graft.rst

File metadata and controls

17 lines (12 loc) · 940 Bytes

GraFT

analysistools-graft (Graph-Filtered Temporal) is a signal extraction method for spatio-temporal data. GraFT uses a diffusion map to learn a graph over spatial pixels that enables for stochastic filtering of learned sparse representations over each pixel's time-trace. The sparse representations are modeled as in a hierarchical dictionary learning framework with correlated decompositions over the graph. GitHub repo <https://github.com/adamshch/GraFT-analysis> NWB tutorials <https://github.com/adamshch/GraFT-analysis/tree/main/code/nwb_tutorials> Cite <https://pubmed.ncbi.nlm.nih.gov/35533160/>

GitHub Repo stars for GraFT