Details
| Original language | English |
|---|---|
| Title of host publication | SIGIR 2022 |
| Subtitle of host publication | Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval |
| Pages | 3219-3223 |
| Number of pages | 5 |
| ISBN (electronic) | 9781450387323 |
| Publication status | Published - 7 Jul 2022 |
| Event | 45th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2022 - Madrid, Spain Duration: 11 Jul 2022 → 15 Jul 2022 |
Abstract
We introduce SparCAssist, a general-purpose risk assessment tool for the machine learning models trained for language tasks. It evaluates models' risk by inspecting their behavior on counterfactuals, namely out-of-distribution instances generated based on the given data instance. The counterfactuals are generated by replacing tokens in rational subsequences identified by ExPred, while the replacements are retrieved using HotFlip or the Masked-Language-Model-based algorithms. The main purpose of our system is to help the human annotators to assess the model's risk on deployment. The counterfactual instances generated during the assessment are the by-product and can be used to train more robust NLP models in the future.
Keywords
- counterfactual interpretation, data-annotation tools, human-in-the-loop machine learning, interpretable machine learning
ASJC Scopus subject areas
- Computer Science(all)
- Computer Graphics and Computer-Aided Design
- Computer Science(all)
- Information Systems
- Computer Science(all)
- Software
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
SIGIR 2022 : Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval. 2022. p. 3219-3223.
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - SparCAssist
T2 - 45th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2022
AU - Zhang, Zijian
AU - Setty, Vinay
AU - Anand, Avishek
N1 - Funding Information: This work is partially funded by project MIRROR under grant agreement No. 832921 (project MIRROR from the European Commission: Migration-Related Risks caused by misconceptions of Opportunities and Requirement) and project ROXANNE, the European Union’s Horizon 2020 research and innovation program under grant agreement No. 833635.
PY - 2022/7/7
Y1 - 2022/7/7
N2 - We introduce SparCAssist, a general-purpose risk assessment tool for the machine learning models trained for language tasks. It evaluates models' risk by inspecting their behavior on counterfactuals, namely out-of-distribution instances generated based on the given data instance. The counterfactuals are generated by replacing tokens in rational subsequences identified by ExPred, while the replacements are retrieved using HotFlip or the Masked-Language-Model-based algorithms. The main purpose of our system is to help the human annotators to assess the model's risk on deployment. The counterfactual instances generated during the assessment are the by-product and can be used to train more robust NLP models in the future.
AB - We introduce SparCAssist, a general-purpose risk assessment tool for the machine learning models trained for language tasks. It evaluates models' risk by inspecting their behavior on counterfactuals, namely out-of-distribution instances generated based on the given data instance. The counterfactuals are generated by replacing tokens in rational subsequences identified by ExPred, while the replacements are retrieved using HotFlip or the Masked-Language-Model-based algorithms. The main purpose of our system is to help the human annotators to assess the model's risk on deployment. The counterfactual instances generated during the assessment are the by-product and can be used to train more robust NLP models in the future.
KW - counterfactual interpretation
KW - data-annotation tools
KW - human-in-the-loop machine learning
KW - interpretable machine learning
UR - http://www.scopus.com/inward/record.url?scp=85135007122&partnerID=8YFLogxK
U2 - 10.48550/arXiv.2205.01588
DO - 10.48550/arXiv.2205.01588
M3 - Conference contribution
AN - SCOPUS:85135007122
SP - 3219
EP - 3223
BT - SIGIR 2022
Y2 - 11 July 2022 through 15 July 2022
ER -