Situated Dialogue Learning through Procedural Environment Generation
This work addresses the challenge of improving agent generalization in interactive dialogue systems for fantasy text adventure games, representing an incremental advancement in curriculum learning methods.
The paper tackles the problem of training goal-driven agents to act and speak in situated environments by using procedurally generated curriculums of increasing difficulty, resulting in significantly higher generalization abilities as measured by zero-shot performance on novel quests.
We teach goal-driven agents to interactively act and speak in situated environments by training on generated curriculums. Our agents operate in LIGHT (Urbanek et al. 2019) -- a large-scale crowd-sourced fantasy text adventure game wherein an agent perceives and interacts with the world through textual natural language. Goals in this environment take the form of character-based quests, consisting of personas and motivations. We augment LIGHT by learning to procedurally generate additional novel textual worlds and quests to create a curriculum of steadily increasing difficulty for training agents to achieve such goals. In particular, we measure curriculum difficulty in terms of the rarity of the quest in the original training distribution -- an easier environment is one that is more likely to have been found in the unaugmented dataset. An ablation study shows that this method of learning from the tail of a distribution results in significantly higher generalization abilities as measured by zero-shot performance on never-before-seen quests.