Compositional Generalization in Grounded Language Learning via Induced Model Sparsity

Sam Spilsbury, Alexander Ilin

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review

52 Downloads (Pure)

Abstract

We provide a study of how induced model sparsity can help achieve compositional generalization and better sample efficiency in grounded language learning problems. We consider simple language-conditioned navigation problems in a grid world environment with disentangled observations. We show that standard neural architectures do not always yield compositional generalization. To address this, we design an agent that contains a goal identification module that encourages sparse correlations between words in the instruction and attributes of objects, composing them together to find the goal. The output of the goal identification module is the input to a value iteration network planner. Our agent maintains a high level of performance on goals containing novel combinations of properties even when learning from a handful of demonstrations. We examine the internal representations of our agent and find the correct correspondences between words in its dictionary and attributes in the environment.

Original languageEnglish
Title of host publicationNAACL 2022 - 2022 Conference of the North American Chapter of the Association for Computational Linguistics
Subtitle of host publicationHuman Language Technologies, Proceedings of the Student Research Workshop
PublisherAssociation for Computational Linguistics
Pages143-155
Number of pages13
ISBN (Electronic)978-1-955917-73-5
DOIs
Publication statusPublished - 2022
MoE publication typeA4 Conference publication
EventConference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies: Human Language Technologies - Seattle, United States
Duration: 10 Jul 202215 Jul 2022

Conference

ConferenceConference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies
Abbreviated titleNAACL
Country/TerritoryUnited States
CitySeattle
Period10/07/202215/07/2022

Fingerprint

Dive into the research topics of 'Compositional Generalization in Grounded Language Learning via Induced Model Sparsity'. Together they form a unique fingerprint.

Cite this