Tweeted By @NathanBenaich

📺 Watch François Chollet's talk at #RAAIS2018 on the limitations of deep learning and the need better evaluation metrics, richer models and stronger priors. He proposes building a reusable library of symbolic AI with geometric AI modules. https://t.co/PjrlU6kDIh cc @fchollet

— Nathan Benaich (@NathanBenaich) August 30, 2018