How Important is the Train-Validation Split in Meta-Learning?

         Yu Bai 1 Minshuo Chen 2 Pan Zhou 1 Tuo Zhao 2 Jason D. Lee 3 Sham Kakade 4
                    Huan Wang 1 Caiming Xiong 1

            Abstract                1. Introduction
   Meta-learning aims to perform fast adaptation       Meta-learning, also known as “learning to learn”, has re-
   on a new task through learning a “prior” from       cently emerged as a powerful paradigm for learning to
   multiple existing tasks. A  ...