Meta-Learning for Domain Generalization in Semantic Parsing
This addresses the challenge of building semantic parsers that generalize to new domains, which is incremental as it applies existing meta-learning techniques to a specific bottleneck in parsing.
The paper tackles the problem of domain generalization in semantic parsing by proposing a meta-learning framework that simulates zero-shot parsing to improve performance on unseen domains, showing significant boosts on Spider and Chinese Spider datasets.
The importance of building semantic parsers which can be applied to new domains and generate programs unseen at training has long been acknowledged, and datasets testing out-of-domain performance are becoming increasingly available. However, little or no attention has been devoted to learning algorithms or objectives which promote domain generalization, with virtually all existing approaches relying on standard supervised learning. In this work, we use a meta-learning framework which targets zero-shot domain generalization for semantic parsing. We apply a model-agnostic training algorithm that simulates zero-shot parsing by constructing virtual train and test sets from disjoint domains. The learning objective capitalizes on the intuition that gradient steps that improve source-domain performance should also improve target-domain performance, thus encouraging a parser to generalize to unseen target domains. Experimental results on the (English) Spider and Chinese Spider datasets show that the meta-learning objective significantly boosts the performance of a baseline parser.