Few weeks ago MIM Solutions team got a cool news that one of our paper was accepted to International Conference on Machine Learning (ICML 2021). The paper has an interesting story behind which would not happen if we did not keep our minds open. More than a year ago, we were working on a more general solution to a parametric flow problem. Previously only very special cases have been solved efficiently. At first sight this does not seem to have anything in common with decomposable submodular function minimization. However, later on we realized that this was the missing step in reducing the problem to the classical flow problem, and that all the new algorithms for maximum flow directly apply to the much more general and useful problem. You can see our paper here.