BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Meta-reinforcement learning - Kris Jensen and Calvin Kao (Universi
 ty of Cambridge)
DTSTART:20210113T110000Z
DTEND:20210113T123000Z
UID:TALK153910@talks.cam.ac.uk
CONTACT:Elre Oldewage
DESCRIPTION:Meta learning allows for generalisation across tasks and has b
 ecome\nincreasingly relevant as machine learning systems are asked to solv
 e\nheterogeneous problems efficiently with less training data. In recent\n
 years\, meta learning has been applied in the context of reinforcement\nle
 arning to build agents that learn to generalise across a distribution\nof 
 Markov decision problems. In this reading group\, we will briefly\nintrodu
 ce the basics of meta reinforcement learning\, cover different\napproaches
  to the problem\, and discuss their uses and limitations. We\nwill also co
 nsider how they compare to more traditional algorithms\, both\nlearned and
  hand-crafted.\n\nRecommended reading:\n\n- Wang et al. 2016 (https://arxi
 v.org/abs/1611.05763) OR Duan et al. 2016\n(https://arxiv.org/abs/1611.027
 79).\n\n- Finn et al. 2017 (https://arxiv.org/abs/1703.03400).\nNagabandi 
 et al. 2019 (https://arxiv.org/abs/1803.11347).\n\n- This blog post also p
 rovides an overview of several of the topics we\nwill cover: https://lilia
 nweng.github.io/lil-log/2019/06/23/meta-reinforcement-learning.html
LOCATION:https://eng-cam.zoom.us/j/86068703738?pwd=YnFleXFQOE1qR1h6Vmtwbno
 0LzFHdz09
END:VEVENT
END:VCALENDAR