Paper accepted at ICML 2021!!

Wonderful news! Our paper “Graph Mixture Density Networks” has been accepted at ICML 2021! Shout out to my supervisors Alessio Micheli and Davide Bacciu that made this possible. We study the problem of learning multi-modal output distributions conditioned on arbitrary input graphs. With GMDN, we can predict if there’s more than one likely outcome associated with an input graph: this is especially useful, for example, when predicting the final outcome of a pandemic when the social network is known. We will release the camera ready very soon!