Transfer of Knowledge in a Reinforcement Learning Setting for a Complex Environment - Progressive Networks in StarCraft II

Jensen, Malthe Dahl ; Madsen, Kaare Bak Toxværd ; Aaes, Andi Rosengreen Kjærsig

4. term

Software, Master

2018

2018-06-07

84

This project is a master thesis by a group on the 10th semester of the software education at Aalborg University. The topic of this project surrounds using reinforcement- and transfer-learning on the complex environment of Starcraft II. We test a number of dierent agent architectures to find a candidate best suited for applying transfer learning. To test if transfer is possible on Starcraft II, we use a network architecture proposed by Google DeepMind in 2016 called progressive networks, which allows us to leverage knowledge from multiple tasks when training on new tasks. At the same time progressive networks do not suer from catastrophic forgetting, which allows us to approximate how much transfer is happening and where in the network it is occurring.

Reinforcement learning ; transfer learning ; A3C ; progressive networks ; starcraft 2 ; starcraft II ; starcraft ; cartpole ; sonic ; sonic the hedgehog ; transfer analysis ; Average Fisher Sensitivity ; actor-critic ; neural networks ; machine learning

Download
View record in AAU Student Projects