Skip to content
This repository has been archived by the owner on Mar 27, 2022. It is now read-only.

On First-Order Meta-Learning Algorithms #110

Open
tm4roon opened this issue Aug 1, 2020 · 0 comments
Open

On First-Order Meta-Learning Algorithms #110

tm4roon opened this issue Aug 1, 2020 · 0 comments
Labels
Meta-Learning Meta-Learning
Projects

Comments

@tm4roon
Copy link
Owner

tm4roon commented Aug 1, 2020

On First-Order Meta-Learning Algorithms

Model-Agnostic Meta-Learning (MAML)は、二次勾配の計算が含まれることからメモリ効率が悪く計算コストが高いことが知られている。本論文では、MAMLの一次近似手法First-Order MAML (FOMAML)を拡張したReptileという手法を提案し、高速かつ頑健に動作することを示していた。手法は以下の通りで非常にシンプルである。

文献情報

参考文献

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
Meta-Learning Meta-Learning
Projects
arXiv
2018
Development

No branches or pull requests

1 participant