Understanding generalization requires rethinking deep learning?

Yamini Bansal, Gal Kaplun, and Boaz Barak (See also paper on arxiv, code on gitlab, upcoming talk by Yamini&Boaz, video of past talk) A central puzzle of deep learning is the question of generalization. In other words, what can we deduce from the training performance of a neural network about its test performance on fresh … Continue reading Understanding generalization requires rethinking deep learning?