SPINN -- an interpretable sparse NN architecture for PDEs

Presented as part of CMInDS seminar series on October 13, 2021 at IIT Bombay. This talk introduces my recent work, carried out in collaboration with my colleague Prabhu Ramachandran, on a class of interpretable neural network architectures for solving PDEs. Link to the published paper can be found below. DOI of JCP paper: https://doi.org/10.1016/j.jcp.2021.11... Arxiv preprint: https://arxiv.org/abs/2102.13037 Disclaimer: All opinions stated here are my own and do not reflect that of any other person or organization. In particular, I'm solely responsible for any misstatements / errors.