Relate to Predict: Towards Task-Independent Knowledge Representations for Reinforcement Learning
This work addresses the challenge of task-independent knowledge representation for RL agents, offering incremental improvements in interpretability and generalization for AI systems.
The paper tackled the problem of interpreting and reusing knowledge in reinforcement learning by introducing an object-centered approach that separates semantic representations and dynamics knowledge, showing that explicit knowledge separation leads to faster learning, better accuracy, generalization, and interpretability in puzzle-like tasks.
Reinforcement Learning (RL) can enable agents to learn complex tasks. However, it is difficult to interpret the knowledge and reuse it across tasks. Inductive biases can address such issues by explicitly providing generic yet useful decomposition that is otherwise difficult or expensive to learn implicitly. For example, object-centered approaches decompose a high dimensional observation into individual objects. Expanding on this, we utilize an inductive bias for explicit object-centered knowledge separation that provides further decomposition into semantic representations and dynamics knowledge. For this, we introduce a semantic module that predicts an objects' semantic state based on its context. The resulting affordance-like object state can then be used to enrich perceptual object representations. With a minimal setup and an environment that enables puzzle-like tasks, we demonstrate the feasibility and benefits of this approach. Specifically, we compare three different methods of integrating semantic representations into a model-based RL architecture. Our experiments show that the degree of explicitness in knowledge separation correlates with faster learning, better accuracy, better generalization, and better interpretability.