Obtaining faithful interpretations from compositional neural networks

Sanjay Subramanian, Ben Bogin, Nitish Gupta, Tomer Wolfson, Sameer Singh, Jonathan Berant, Matt Gardner

פרסום מחקרי: פרק בספר / בדוח / בכנספרסום בספר כנסביקורת עמיתים

תקציר

Neural module networks (NMNs) are a popular approach for modeling compositionality: they achieve high accuracy when applied to problems in language and vision, while reflecting the compositional structure of the problem in the network architecture. However, prior work implicitly assumed that the structure of the network modules, describing the abstract reasoning process, provides a faithful explanation of the model's reasoning; that is, that all modules perform their intended behaviour. In this work, we propose and conduct a systematic evaluation of the intermediate outputs of NMNs on NLVR2 and DROP, two datasets which require composing multiple reasoning steps. We find that the intermediate outputs differ from the expected output, illustrating that the network structure does not provide a faithful explanation of model behaviour. To remedy that, we train the model with auxiliary supervision and propose particular choices for module architecture that yield much better faithfulness, at a minimal cost to accuracy.

שפה מקוריתאנגלית
כותר פרסום המארחACL 2020 - 58th Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference
מוציא לאורAssociation for Computational Linguistics (ACL)
עמודים5594-5608
מספר עמודים15
מסת"ב (אלקטרוני)9781952148255
סטטוס פרסוםפורסם - 2020
אירוע58th Annual Meeting of the Association for Computational Linguistics, ACL 2020 - Virtual, Online, ארצות הברית
משך הזמן: 5 יולי 202010 יולי 2020

סדרות פרסומים

שםProceedings of the Annual Meeting of the Association for Computational Linguistics

כנס

כנס58th Annual Meeting of the Association for Computational Linguistics, ACL 2020
מדינה/אזורארצות הברית
עירVirtual, Online
תקופה5/07/2010/07/20

ASJC Scopus subject areas

  • ???subjectarea.asjc.1700.1706???
  • ???subjectarea.asjc.3300.3310???
  • ???subjectarea.asjc.1200.1203???

טביעת אצבע

להלן מוצגים תחומי המחקר של הפרסום 'Obtaining faithful interpretations from compositional neural networks'. יחד הם יוצרים טביעת אצבע ייחודית.

פורמט ציטוט ביבליוגרפי