Systematic Generalisation through Task Temporal Logic and Deep Reinforcement Learning
This addresses the challenge of enabling AI agents to generalize to unseen tasks in domains requiring formal specifications, representing an incremental advance in neuro-symbolic methods.
The paper tackles the problem of achieving systematic zero-shot generalization of formally specified instructions by introducing a neuro-symbolic agent that combines deep reinforcement learning with temporal logic, finding that convolutional layer architecture is key and that systematic learning can emerge with abstract operators like negation from few examples.
This work introduces a neuro-symbolic agent that combines deep reinforcement learning (DRL) with temporal logic (TL) to achieve systematic zero-shot, i.e., never-seen-before, generalisation of formally specified instructions. In particular, we present a neuro-symbolic framework where a symbolic module transforms TL specifications into a form that helps the training of a DRL agent targeting generalisation, while a neural module learns systematically to solve the given tasks. We study the emergence of systematic learning in different settings and find that the architecture of the convolutional layers is key when generalising to new instructions. We also provide evidence that systematic learning can emerge with abstract operators such as negation when learning from a few training examples, which previous research have struggled with.