Timezone: »

Provable Generalization of Overparameterized Meta-learning Trained with SGD
Yu Huang · Yingbin Liang · Longbo Huang


Despite the empirical success of deep meta-learning, theoretical understanding of overparameterized meta-learning is still limited. This paper studies the generalization of a widely used meta-learning approach, Model-Agnostic Meta-Learning (MAML), which aims to find a good initialization for fast adaptation to new tasks. Under a mixed linear regression model, we analyze the generalization properties of MAML trained with SGD in the overparameterized regime. We provide both upper and lower bounds for the excess risk of MAML, which captures how SGD dynamics affect these generalization bounds. With such sharp characterizations, we further explore how various learning parameters impact the generalization capability of overparameterized MAML, including explicitly identifying typical data and task distributions that can achieve diminishing generalization error with overparameterization, and characterizing the impact of adaptation learning rate on both excess risk and the early stopping time. Our theoretical findings are further validated by experiments.

Author Information

Yu Huang (Institute for Interdisciplinary Information Sciences (IIIS), Tsinghua University)
Yingbin Liang (The Ohio State University)
Longbo Huang (IIIS, Tsinghua Univeristy)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors