Learning to Learn

注:本文内容来自下述链接以及自己补充的近期的相关工作。


  • 2018-07-28 最前沿:Meta Learning前沿进展扫描
  • 2018-06-02 任务无偏的元学习(Task-Agnostic):最小化任务性能之间的不平等
  • 2017-08-21 最前沿:百家争鸣的Meta Learning/Learning to learn

Meta Learning从结合角度上看可以分成三大块:Supervised Learning Unsupervised Learning/Generative Model以及Reinforcement Learning。那么本文仅探讨Meta Learning在Supervised Learning上的应用,其实也就是Few Shot Learning。

State of the Art

截图自Meta-Learning with Latent Embedding Optimization

你可能感兴趣的:(Learning to Learn)