Active Exploration for Learning Symbolic Representations
This work addresses the challenge of efficient symbolic representation learning for agents in simulated environments, though it appears incremental as it builds on existing exploration methods.
The authors tackled the problem of data-efficient learning of abstract symbolic models in environments by introducing an online active exploration algorithm that guides exploration based on model uncertainty. They showed that their algorithm outperforms random and greedy exploration policies on two computer game domains, with specific performance gains reported in the abstract.
We introduce an online active exploration algorithm for data-efficiently learning an abstract symbolic model of an environment. Our algorithm is divided into two parts: the first part quickly generates an intermediate Bayesian symbolic model from the data that the agent has collected so far, which the agent can then use along with the second part to guide its future exploration towards regions of the state space that the model is uncertain about. We show that our algorithm outperforms random and greedy exploration policies on two different computer game domains. The first domain is an Asteroids-inspired game with complex dynamics but basic logical structure. The second is the Treasure Game, with simpler dynamics but more complex logical structure.