no code implementations • 11 Dec 2019 • Kadir Gümüs, Alex Alvarado, Bin Chen, Christian Häger, Erik Agrell
GMI-based end-to-end learning is shown to be highly nonconvex.