Meta-learning — Introduction and overview

21/05/2020

Speaker: Karol Arndt

Robotics Seminar Series. Fifth Session – 22nd May 2020, 13:00-14:00, via zoom. Link to event: https://aalto.zoom.us/j/62124942899

Abstract:

In recent years, meta learning has seen increasing attention in the machine learning and robotics communities. In this presentation, I will provide an introduction and overview to meta-learning techniques: what problems they’re trying to solve, what basic algorithms can be used to solve these problems, and how the existing methods can be applied in the context of reinforcement learning.

Reference:
  • C. Finn et al., Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks, 2017, https://arxiv.org/abs/1703.03400
  • P. Ortega et al., Meta-learning of Sequential Strategies, 2019, https://arxiv.org/abs/1905.03030
  • E. Grant et al, Recasting Gradient-Based Meta-Learning as Hierarchical Bayes, 2018, https://arxiv.org/abs/1801.08930
  • B. Stadie et al., Some Considerations on Learning to Explore via Meta-Reinforcement Learning, 2018, https://arxiv.org/abs/1803.011
Tags: