Discovering Blind Spots in Reinforcement Learning
AAMAS, pp. 1017-1025, 2018.
Agents trained in simulation may make errors in the real world due to mismatches between training and execution environments. These mistakes can be dangerous and difficult to discover because the agent cannot predict them a priori. We propose using oracle feedback to learn a predictive model of these blind spots to reduce costly errors in...More
Full Text (Upload PDF)
PPT (Upload PPT)