Effective Meta-Regularization by Kernelized Proximal Regularization

Jiang，Weisen1,2; Kwok，James T.2; Zhang，Yu1,3

2021

1049-5258

Advances in Neural Information Processing Systems

31

26212-26222

We study the problem of meta-learning, which has proved to be advantageous to accelerate learning new tasks with a few samples. The recent approaches based on deep kernels achieve the state-of-the-art performance. However, the regularizers in their base learners are not learnable. In this paper, we propose an algorithm called MetaProx to learn a proximal regularizer for the base learner. We theoretically establish the convergence of MetaProx. Experimental results confirm the advantage of the proposed algorithm.

First ; Corresponding

English

EI

National Natural Science Foundation of China[62076118];

20222512238272

2-s2.0-85131910909

Scopus

1.Department of Computer Science and Engineering,Southern University of Science and Technology,China
2.Department of Computer Science and Engineering,Hong Kong University of Science and Technology,Hong Kong
3.Peng Cheng Laboratory,China

Jiang，Weisen,Kwok，James T.,Zhang，Yu. Effective Meta-Regularization by Kernelized Proximal Regularization[C],2021:26212-26222.