Trautner, Margolis, and Ravela  won the "Top 5" recognition for the paper presented at the  DDDAS 2020 conference. This paper contains three fundamentally transformative components: 1) Informative Ensemble Learning provides an efficacious parallel distributed path to learning and can replace Backprop 2) Information Gain provides a framework for optimizing $\ell_0$ problems in a Greedy manner, which appears to converge faster than $\ell_1$ approximation. 3)  Equations can be learned from data to build stable hybrid dynamical systems. 

Please click for  Informative Neural Ensemble Kalman Learning paper

Watch the video: