Cross-functional Analysis of Generalisation in Behavioural Learning ...
In behavioural testing, system functionalities underrepresented in the standard evaluation setting (with a held-out test set) are validated through controlled input-output pairs. Optimising performance on the behavioural tests during training (behavioural learning) would improve coverage of phenomen...
Main Authors: | , |
---|---|
Format: | Text |
Language: | unknown |
Published: |
arXiv
2023
|
Subjects: | |
Online Access: | https://dx.doi.org/10.48550/arxiv.2305.12951 https://arxiv.org/abs/2305.12951 |
id |
ftdatacite:10.48550/arxiv.2305.12951 |
---|---|
record_format |
openpolar |
spelling |
ftdatacite:10.48550/arxiv.2305.12951 2023-10-01T03:55:03+02:00 Cross-functional Analysis of Generalisation in Behavioural Learning ... de Araujo, Pedro Henrique Luz Roth, Benjamin 2023 https://dx.doi.org/10.48550/arxiv.2305.12951 https://arxiv.org/abs/2305.12951 unknown arXiv https://dx.doi.org/10.1162/tacl_a_00590 Creative Commons Attribution 4.0 International https://creativecommons.org/licenses/by/4.0/legalcode cc-by-4.0 Computation and Language cs.CL Machine Learning cs.LG FOS Computer and information sciences ScholarlyArticle Article article-journal Text 2023 ftdatacite https://doi.org/10.48550/arxiv.2305.1295110.1162/tacl_a_00590 2023-09-04T15:13:49Z In behavioural testing, system functionalities underrepresented in the standard evaluation setting (with a held-out test set) are validated through controlled input-output pairs. Optimising performance on the behavioural tests during training (behavioural learning) would improve coverage of phenomena not sufficiently represented in the i.i.d. data and could lead to seemingly more robust models. However, there is the risk that the model narrowly captures spurious correlations from the behavioural test suite, leading to overestimation and misrepresentation of model performance -- one of the original pitfalls of traditional evaluation. In this work, we introduce BeLUGA, an analysis method for evaluating behavioural learning considering generalisation across dimensions of different granularity levels. We optimise behaviour-specific loss functions and evaluate models on several partitions of the behavioural test suite controlled to leave out specific phenomena. An aggregate score measures generalisation to unseen ... : 16 pages, 1 figure. To be published in the Transactions of the Association for Computational Linguistics (TACL). This preprint is a pre-MIT Press publication version ... Text Beluga Beluga* DataCite Metadata Store (German National Library of Science and Technology) |
institution |
Open Polar |
collection |
DataCite Metadata Store (German National Library of Science and Technology) |
op_collection_id |
ftdatacite |
language |
unknown |
topic |
Computation and Language cs.CL Machine Learning cs.LG FOS Computer and information sciences |
spellingShingle |
Computation and Language cs.CL Machine Learning cs.LG FOS Computer and information sciences de Araujo, Pedro Henrique Luz Roth, Benjamin Cross-functional Analysis of Generalisation in Behavioural Learning ... |
topic_facet |
Computation and Language cs.CL Machine Learning cs.LG FOS Computer and information sciences |
description |
In behavioural testing, system functionalities underrepresented in the standard evaluation setting (with a held-out test set) are validated through controlled input-output pairs. Optimising performance on the behavioural tests during training (behavioural learning) would improve coverage of phenomena not sufficiently represented in the i.i.d. data and could lead to seemingly more robust models. However, there is the risk that the model narrowly captures spurious correlations from the behavioural test suite, leading to overestimation and misrepresentation of model performance -- one of the original pitfalls of traditional evaluation. In this work, we introduce BeLUGA, an analysis method for evaluating behavioural learning considering generalisation across dimensions of different granularity levels. We optimise behaviour-specific loss functions and evaluate models on several partitions of the behavioural test suite controlled to leave out specific phenomena. An aggregate score measures generalisation to unseen ... : 16 pages, 1 figure. To be published in the Transactions of the Association for Computational Linguistics (TACL). This preprint is a pre-MIT Press publication version ... |
format |
Text |
author |
de Araujo, Pedro Henrique Luz Roth, Benjamin |
author_facet |
de Araujo, Pedro Henrique Luz Roth, Benjamin |
author_sort |
de Araujo, Pedro Henrique Luz |
title |
Cross-functional Analysis of Generalisation in Behavioural Learning ... |
title_short |
Cross-functional Analysis of Generalisation in Behavioural Learning ... |
title_full |
Cross-functional Analysis of Generalisation in Behavioural Learning ... |
title_fullStr |
Cross-functional Analysis of Generalisation in Behavioural Learning ... |
title_full_unstemmed |
Cross-functional Analysis of Generalisation in Behavioural Learning ... |
title_sort |
cross-functional analysis of generalisation in behavioural learning ... |
publisher |
arXiv |
publishDate |
2023 |
url |
https://dx.doi.org/10.48550/arxiv.2305.12951 https://arxiv.org/abs/2305.12951 |
genre |
Beluga Beluga* |
genre_facet |
Beluga Beluga* |
op_relation |
https://dx.doi.org/10.1162/tacl_a_00590 |
op_rights |
Creative Commons Attribution 4.0 International https://creativecommons.org/licenses/by/4.0/legalcode cc-by-4.0 |
op_doi |
https://doi.org/10.48550/arxiv.2305.1295110.1162/tacl_a_00590 |
_version_ |
1778523175431176192 |