Skip to yearly menu bar Skip to main content


Align-RUDDER: Learning From Few Demonstrations by Reward Redistribution

Vihang Patil · Markus Hofmarcher · Marius-Constantin Dinu · Matthias Dorfer · Patrick Blies · Johannes Brandstetter · Jose A. Arjona-Medina · Sepp Hochreiter

Hall E #826

Keywords: [ RL: Deep RL ] [ RL: Total Cost/Reward ] [ Reinforcement Learning ]


Reinforcement learning algorithms require many samples when solving complex hierarchical tasks with sparse and delayed rewards. For such complex tasks, the recently proposed RUDDER uses reward redistribution to leverage steps in the Q-function that are associated with accomplishing sub-tasks. However, often only few episodes with high rewards are available as demonstrations since current exploration strategies cannot discover them in reasonable time. In this work, we introduce Align-RUDDER, which utilizes a profile model for reward redistribution that is obtained from multiple sequence alignment of demonstrations. Consequently, Align-RUDDER employs reward redistribution effectively and, thereby, drastically improves learning on few demonstrations. Align-RUDDER outperforms competitors on complex artificial tasks with delayed rewards and few demonstrations. On the Minecraft ObtainDiamond task, Align-RUDDER is able to mine a diamond, though not frequently. Code is available at

Chat is not available.