RL — Meta-Learning

  1. Sample efficiency: Deep learning has poor sample efficiency. For example, to recognize digit handwriting, we typically read 6000 samples per digit.
  2. Poor transferability. We don’t learn from previous experience or knowledge.
  • recurrent models,
  • meta-optimization, and
  • metric learning.

Few-Shot Learning

Meta-learning program Reptile from OpenAI
Mistaken the duck toy as a savings bank or a musical instrument
Modified from source
Source

Recurrent Models

Modified from source
source

Learning optimizers

Source
Source
Source

Metric learning

Source

Other approaches

source

Thoughts

  • Collect better information to learn.
  • Learn from past experiences better.
  • Know better how to represent information.
  • How to optimize (solve) model better.
  • Explore better.
  • Associate better.
  • Generalize better.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store