Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204233172> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W3204233172 abstract "The ability to abstract, count, and use System 2 reasoning are well-known manifestations of intelligence and understanding. In this paper, we argue, using the example of the ``Look and Say puzzle, that although deep neural networks can exhibit high `competence' (as measured by accuracy) when trained on large data sets (2M examples in our case), they do not show any sign on the deeper understanding of the problem, or what D. Dennett calls `comprehension'. We report on two sets experiments on the ``Look and Say puzzle data. We view the problem as building a translator from one set of tokens to another. We apply both standard LSTMs and Transformer/Attention -- based neural networks, using publicly available machine translation software. We observe that despite the amazing accuracy (on both, training and test data), the performance of the trained programs on the actual L&S sequence is bad. We then discuss a few possible ramifications of this finding and connections to other work, experimental and theoretical. First, from the cognitive science perspective, we argue that we need better mathematical models of abstraction. Second, the classical and more recent results on the universality of neural networks should be re-examined for functions acting on discrete data sets. Mapping on discrete sets usually have no natural continuous extensions. This connects the results on a simple puzzle to more sophisticated results on modeling of mathematical functions, where algebraic functions are more difficult to model than e.g. differential equations. Third, we hypothesize that for problems such as ``Look and Say, computing the parity of bitstrings, or learning integer addition, it might be worthwhile to introduce concepts from topology, where continuity is defined without the reference to the concept of distance." @default.
- W3204233172 created "2021-10-11" @default.
- W3204233172 creator A5041770151 @default.
- W3204233172 date "2021-09-27" @default.
- W3204233172 modified "2023-09-27" @default.
- W3204233172 title "Abstraction, Reasoning and Deep Learning: A Study of the Look and Say Sequence." @default.
- W3204233172 cites W1496383623 @default.
- W3204233172 cites W1979013627 @default.
- W3204233172 cites W1988115241 @default.
- W3204233172 cites W2103496339 @default.
- W3204233172 cites W2115733720 @default.
- W3204233172 cites W2167366201 @default.
- W3204233172 cites W2401823607 @default.
- W3204233172 cites W2752099845 @default.
- W3204233172 cites W2771918024 @default.
- W3204233172 cites W2786112702 @default.
- W3204233172 cites W2899011312 @default.
- W3204233172 cites W2962723285 @default.
- W3204233172 cites W2963267799 @default.
- W3204233172 cites W2963305465 @default.
- W3204233172 cites W2970410772 @default.
- W3204233172 cites W2994278562 @default.
- W3204233172 cites W2995273672 @default.
- W3204233172 cites W2996686709 @default.
- W3204233172 cites W2997789497 @default.
- W3204233172 cites W3008152157 @default.
- W3204233172 cites W3013946425 @default.
- W3204233172 cites W3034381049 @default.
- W3204233172 cites W3034830866 @default.
- W3204233172 cites W3034942609 @default.
- W3204233172 cites W3035091181 @default.
- W3204233172 cites W3035314023 @default.
- W3204233172 cites W3091917188 @default.
- W3204233172 cites W3111753701 @default.
- W3204233172 cites W3113024959 @default.
- W3204233172 cites W3122457781 @default.
- W3204233172 cites W3132397641 @default.
- W3204233172 cites W3135593154 @default.
- W3204233172 cites W3140579943 @default.
- W3204233172 cites W3159783455 @default.
- W3204233172 cites W3170954064 @default.
- W3204233172 cites W3178308100 @default.
- W3204233172 cites W3201369838 @default.
- W3204233172 hasPublicationYear "2021" @default.
- W3204233172 type Work @default.
- W3204233172 sameAs 3204233172 @default.
- W3204233172 citedByCount "0" @default.
- W3204233172 crossrefType "posted-content" @default.
- W3204233172 hasAuthorship W3204233172A5041770151 @default.
- W3204233172 hasConcept C111472728 @default.
- W3204233172 hasConcept C119857082 @default.
- W3204233172 hasConcept C124304363 @default.
- W3204233172 hasConcept C138885662 @default.
- W3204233172 hasConcept C154945302 @default.
- W3204233172 hasConcept C41008148 @default.
- W3204233172 hasConcept C50644808 @default.
- W3204233172 hasConcept C80444323 @default.
- W3204233172 hasConceptScore W3204233172C111472728 @default.
- W3204233172 hasConceptScore W3204233172C119857082 @default.
- W3204233172 hasConceptScore W3204233172C124304363 @default.
- W3204233172 hasConceptScore W3204233172C138885662 @default.
- W3204233172 hasConceptScore W3204233172C154945302 @default.
- W3204233172 hasConceptScore W3204233172C41008148 @default.
- W3204233172 hasConceptScore W3204233172C50644808 @default.
- W3204233172 hasConceptScore W3204233172C80444323 @default.
- W3204233172 hasLocation W32042331721 @default.
- W3204233172 hasOpenAccess W3204233172 @default.
- W3204233172 hasPrimaryLocation W32042331721 @default.
- W3204233172 hasRelatedWork W134537301 @default.
- W3204233172 hasRelatedWork W1488273532 @default.
- W3204233172 hasRelatedWork W1597904069 @default.
- W3204233172 hasRelatedWork W2003502238 @default.
- W3204233172 hasRelatedWork W2004652576 @default.
- W3204233172 hasRelatedWork W2070715309 @default.
- W3204233172 hasRelatedWork W2177508269 @default.
- W3204233172 hasRelatedWork W2186009552 @default.
- W3204233172 hasRelatedWork W2405529744 @default.
- W3204233172 hasRelatedWork W2492604755 @default.
- W3204233172 hasRelatedWork W2519116206 @default.
- W3204233172 hasRelatedWork W2741221930 @default.
- W3204233172 hasRelatedWork W27961112 @default.
- W3204233172 hasRelatedWork W2947795716 @default.
- W3204233172 hasRelatedWork W2951295112 @default.
- W3204233172 hasRelatedWork W2964071299 @default.
- W3204233172 hasRelatedWork W2990182607 @default.
- W3204233172 hasRelatedWork W3099579437 @default.
- W3204233172 hasRelatedWork W647012475 @default.
- W3204233172 hasRelatedWork W3021558512 @default.
- W3204233172 isParatext "false" @default.
- W3204233172 isRetracted "false" @default.
- W3204233172 magId "3204233172" @default.
- W3204233172 workType "article" @default.