常识:语言和智能的暗物质(VLDB 2023主题演讲)

IF 2.6 3区计算机科学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS Proceedings of the Vldb Endowment Pub Date : 2023-08-01 DOI:10.14778/3611540.3611638

Yejin Choi

{"title":"常识:语言和智能的暗物质(VLDB 2023主题演讲)","authors":"Yejin Choi","doi":"10.14778/3611540.3611638","DOIUrl":null,"url":null,"abstract":"Scale appears to be the winning recipe in today's leaderboards. And yet, extreme-scale neural models are (un)surprisingly brittle and make errors that are often nonsensical and even counterintuitive. In this talk, I will argue for the importance of knowledge, especially commonsense knowledge, as well as inference-time reasoning algorithms, and demonstrate how smaller models developed in academia can still have an edge over larger industry-scale models, if powered with knowledge and/or reasoning algorithms.","PeriodicalId":54220,"journal":{"name":"Proceedings of the Vldb Endowment","volume":"18 1","pages":"0"},"PeriodicalIF":2.6000,"publicationDate":"2023-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Common Sense: The Dark Matter of Language and Intelligence (VLDB 2023 Keynote)\",\"authors\":\"Yejin Choi\",\"doi\":\"10.14778/3611540.3611638\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Scale appears to be the winning recipe in today's leaderboards. And yet, extreme-scale neural models are (un)surprisingly brittle and make errors that are often nonsensical and even counterintuitive. In this talk, I will argue for the importance of knowledge, especially commonsense knowledge, as well as inference-time reasoning algorithms, and demonstrate how smaller models developed in academia can still have an edge over larger industry-scale models, if powered with knowledge and/or reasoning algorithms.\",\"PeriodicalId\":54220,\"journal\":{\"name\":\"Proceedings of the Vldb Endowment\",\"volume\":\"18 1\",\"pages\":\"0\"},\"PeriodicalIF\":2.6000,\"publicationDate\":\"2023-08-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Proceedings of the Vldb Endowment\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.14778/3611540.3611638\",\"RegionNum\":3,\"RegionCategory\":\"计算机科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"COMPUTER SCIENCE, INFORMATION SYSTEMS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the Vldb Endowment","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.14778/3611540.3611638","RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"COMPUTER SCIENCE, INFORMATION SYSTEMS","Score":null,"Total":0}

引用次数: 0

摘要

规模似乎是当今排行榜的制胜秘诀。然而，极端尺度的神经模型非常脆弱，经常会犯一些荒谬甚至违反直觉的错误。在这次演讲中，我将论证知识的重要性，尤其是常识性知识，以及推理时间推理算法，并展示学术界开发的小型模型如何仍然比大型工业规模的模型具有优势，如果有知识和/或推理算法的支持。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

Common Sense: The Dark Matter of Language and Intelligence (VLDB 2023 Keynote)

Scale appears to be the winning recipe in today's leaderboards. And yet, extreme-scale neural models are (un)surprisingly brittle and make errors that are often nonsensical and even counterintuitive. In this talk, I will argue for the importance of knowledge, especially commonsense knowledge, as well as inference-time reasoning algorithms, and demonstrate how smaller models developed in academia can still have an edge over larger industry-scale models, if powered with knowledge and/or reasoning algorithms.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

Proceedings of the Vldb Endowment Computer Science-General Computer Science

CiteScore

7.70

自引率

0.00%

发文量

期刊介绍： The Proceedings of the VLDB (PVLDB) welcomes original research papers on a broad range of research topics related to all aspects of data management, where systems issues play a significant role, such as data management system technology and information management infrastructures, including their very large scale of experimentation, novel architectures, and demanding applications as well as their underpinning theory. The scope of a submission for PVLDB is also described by the subject areas given below. Moreover, the scope of PVLDB is restricted to scientific areas that are covered by the combined expertise on the submission’s topic of the journal’s editorial board. Finally, the submission’s contributions should build on work already published in data management outlets, e.g., PVLDB, VLDBJ, ACM SIGMOD, IEEE ICDE, EDBT, ACM TODS, IEEE TKDE, and go beyond a syntactic citation.