TY - JOUR
T1 - Bayesian model-agnostic meta-learning
AU - Yoon, Jaesik
AU - Kim, Taesup
AU - Dia, Ousmane
AU - Kim, Sungwoong
AU - Bengio, Yoshua
AU - Ahn, Sungjin
N1 - Funding Information:
JY thanks SAP and Kakao Brain for their support. TK thanks NSERC, MILA and Kakao Brain for their support. YB thanks CIFAR, NSERC, IBM, Google, Facebook and Microsoft for their support. SA, Element AI Fellow, thanks Nicolas Chapados, Pedro Oliveira Pinheiro, Alexandre Lacoste, Negar Rostamzadeh for helpful discussions and feedback.
PY - 2018
Y1 - 2018
N2 - Due to the inherent model uncertainty, learning to infer Bayesian posterior from a few-shot dataset is an important step towards robust meta-learning. In this paper, we propose a novel Bayesian model-agnostic meta-learning method. The proposed method combines efficient gradient-based meta-learning with nonparametric variational inference in a principled probabilistic framework. Unlike previous methods, during fast adaptation, the method is capable of learning complex uncertainty structure beyond a simple Gaussian approximation, and during meta-update, a novel Bayesian mechanism prevents meta-level overfitting. Remaining a gradient-based method, it is also the first Bayesian model-agnostic meta-learning method applicable to various tasks including reinforcement learning. Experiment results show the accuracy and robustness of the proposed method in sinusoidal regression, image classification, active learning, and reinforcement learning.
AB - Due to the inherent model uncertainty, learning to infer Bayesian posterior from a few-shot dataset is an important step towards robust meta-learning. In this paper, we propose a novel Bayesian model-agnostic meta-learning method. The proposed method combines efficient gradient-based meta-learning with nonparametric variational inference in a principled probabilistic framework. Unlike previous methods, during fast adaptation, the method is capable of learning complex uncertainty structure beyond a simple Gaussian approximation, and during meta-update, a novel Bayesian mechanism prevents meta-level overfitting. Remaining a gradient-based method, it is also the first Bayesian model-agnostic meta-learning method applicable to various tasks including reinforcement learning. Experiment results show the accuracy and robustness of the proposed method in sinusoidal regression, image classification, active learning, and reinforcement learning.
UR - http://www.scopus.com/inward/record.url?scp=85064441999&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85064441999&partnerID=8YFLogxK
M3 - Conference article
AN - SCOPUS:85064441999
SN - 1049-5258
VL - 2018-December
SP - 7332
EP - 7342
JO - Advances in Neural Information Processing Systems
JF - Advances in Neural Information Processing Systems
T2 - 32nd Conference on Neural Information Processing Systems, NeurIPS 2018
Y2 - 2 December 2018 through 8 December 2018
ER -