Fitted Q-Learning for Relational Domains
This work addresses the challenge of scaling reinforcement learning to relational domains, which is incremental as it adapts existing fitted Q-learning methods to a new context.
The paper tackled the problem of Approximate Dynamic Programming in relational domains by developing the first relational fitted Q-learning algorithms, showing that the framework performs reasonably well on standard domains without using domain models and with fewer training trajectories.
We consider the problem of Approximate Dynamic Programming in relational domains. Inspired by the success of fitted Q-learning methods in propositional settings, we develop the first relational fitted Q-learning algorithms by representing the value function and Bellman residuals. When we fit the Q-functions, we show how the two steps of Bellman operator; application and projection steps can be performed using a gradient-boosting technique. Our proposed framework performs reasonably well on standard domains without using domain models and using fewer training trajectories.