Latent Compositional Representations Improve Systematic Generalization in Grounded Question Answering
Published
2022-01-04
Ben Bogin
,
Sanjay Subramanian
,
Matt Gardner
,
Jonathan Berant
Ben Bogin
Tel Aviv University
Sanjay Subramanian
Allen Institute for AI
Matt Gardner
Allen Institute for AI
Jonathan Berant
Tel Aviv University, Allen Institute for AI
Abstract
Answering questions that involve multi-step reasoning requires decomposing them and using the answers of intermediate steps to reach the final answer. However, state-of-the-art models in grounded question answering often do not explicitly perform decomposition, leading to difficulties in generalization to out-of-distribution examples. In this work, we propose a model that computes a representation and denotation for all question spans in a bottom-up, compositional manner using a CKY-style parser. Our model induces latent trees, driven by end-to-end (the answer) supervision only. We show that this inductive bias towards tree structures dramatically improves systematic generalization to out-of-distribution examples, compared to strong baselines on an arithmetic expressions benchmark as well as on CLOSURE, a dataset that focuses on systematic generalization for grounded question answering. On this challenging dataset, our model reaches an accuracy of 96.1%, significantly higher than prior models that almost perfectly solve the task on a random, in-distribution split.
Presented at NAACL 2021
Article at MIT Press