New paper from Open AI that uses meta-learning to do hierarchical reinforcement learning. This could be very useful for complicated biological/chemical problems where action spaces are large. Plus, this is a nice go-between building on dc.rl and dc.metalearning. Would be awesome to get an implementation in DeepChem.
New paper from Open AI that uses meta-learning to do hierarchical reinforcement learning. This could be very useful for complicated biological/chemical problems where action spaces are large. Plus, this is a nice go-between building on
dc.rl
anddc.metalearning
. Would be awesome to get an implementation in DeepChem.https://blog.openai.com/learning-a-hierarchy/
H/T @cxhernandez