Deep Learning Alchemy

Simon discusses the lack of overarching theories in deep learning, highlighting the gap between experimental findings and theoretical explanations. He delves into the challenges of relying on unrealistic assumptions and the disconnect between empirical results and theoretical predictions.