{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/learning-to-make-analogies-by-contrasting","title":"Learning to Make Analogies by Contrasting Abstract Relational Structure","arxiv_id":"1902.00120","date":"2019-01-31","proceeding":"ICLR 2019 5","authors":["Felix Hill","Adam Santoro","David G. T. Barrett","Ari S. Morcos","Timothy Lillicrap"],"abstract":"Analogical reasoning has been a principal focus of various waves of AI\nresearch. Analogy is particularly challenging for machines because it requires\nrelational structures to be represented such that they can be flexibly applied\nacross diverse domains of experience. Here, we study how analogical reasoning\ncan be induced in neural networks that learn to perceive and reason about raw\nvisual data. We find that the critical factor for inducing such a capacity is\nnot an elaborate architecture, but rather, careful attention to the choice of\ndata and the manner in which it is presented to the model. The most robust\ncapacity for analogical reasoning is induced when networks learn analogies by\ncontrasting abstract relational structures in their input domains, a training\nmethod that uses only the input data to force models to learn about important\nabstract features. Using this technique we demonstrate capacities for complex,\nvisual and symbolic analogy making and generalisation in even the simplest\nneural network architectures.","url_abs":"http://arxiv.org/abs/1902.00120v1","url_pdf":"http://arxiv.org/pdf/1902.00120v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"learning-to-make-analogies-by-contrasting","repo_url":"https://github.com/deepmind/abstract-reasoning-matrices","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"learning-to-make-analogies-by-contrasting","repo_url":"https://github.com/taylorwwebb/learning_representations_that_support_extrapolation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1902.00120","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}