Neural-Guided, Bidirectional Program Search for Abstraction and Reasoning

Simon Alford, Anshula Gandhi, Akshay Rangamani, Andrzej Banburski, Tony Wang, Sylee Dandekar, John Chin, Tomaso Poggio, Peter Chin

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Scopus citations

Abstract

One of the challenges facing artificial intelligence research today is designing systems capable of utilizing systematic reasoning to generalize to new tasks. The Abstraction and Reasoning Corpus (ARC) measures such a capability through a set of visual reasoning tasks. In this paper we report incremental progress on ARC and lay the foundations for two approaches to abstraction and reasoning not based in brute-force search. We first apply an existing program synthesis system called DreamCoder to create symbolic abstractions out of tasks solved so far, and show how it enables solving of progressively more challenging ARC tasks. Second, we design a reasoning algorithm motivated by the way humans approach ARC. Our algorithm constructs a search graph and reasons over this graph structure to discover task solutions. More specifically, we extend existing execution-guided program synthesis approaches with deductive reasoning based on function inverse semantics to enable a neural-guided bidirectional search algorithm. We demonstrate the effectiveness of the algorithm on three domains: ARC, 24-Game tasks, and a ‘double-and-add’ arithmetic puzzle.

Original languageEnglish (US)
Title of host publicationComplex Networks and Their Applications X - Proceedings of the 10th International Conference on Complex Networks and Their Applications COMPLEX NETWORKS 2021
EditorsRosa Maria Benito, Chantal Cherifi, Hocine Cherifi, Esteban Moro, Luis M. Rocha, Marta Sales-Pardo
PublisherSpringer Science and Business Media Deutschland GmbH
Pages657-668
Number of pages12
ISBN (Print)9783030934088
DOIs
StatePublished - 2022
Externally publishedYes
Event10th International Conference on Complex Networks and Their Applications, COMPLEX NETWORKS 2021 - Madrid, Spain
Duration: Nov 30 2021Dec 2 2021

Publication series

NameStudies in Computational Intelligence
Volume1015
ISSN (Print)1860-949X
ISSN (Electronic)1860-9503

Conference

Conference10th International Conference on Complex Networks and Their Applications, COMPLEX NETWORKS 2021
Country/TerritorySpain
CityMadrid
Period11/30/2112/2/21

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence

Keywords

  • Abstraction
  • Neural networks
  • Program synthesis
  • Reasoning

Fingerprint

Dive into the research topics of 'Neural-Guided, Bidirectional Program Search for Abstraction and Reasoning'. Together they form a unique fingerprint.

Cite this