AbstainGNN: Teaching Graph Neural Networks to Abstain for Graph Classification Conference: Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2; August 09–13, 2026; Jeju Island, Republic of Korea Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2 (KDD ’26), August 09–13, 2026, Jeju Island, Republic of Korea CCS: Computing methodologies Machine learning CCS: Information systems Data mining
Xixun Lin, Zhiheng Zhou, Zhengyin Zhang, Yancheng Chen, Shuai Zhang, Ge Zhang, Shichao Zhu, Lixin Zou, Chuan Zhou, Peng Zhang, Shirui Pan, Yanan Cao · 2026
Read AbstainGNN: Teaching Graph Neural Networks to Abstain for Graph Classification Conference: Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2; August 09–13, 2026; Jeju Island, Republic of Korea Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2 (KDD ’26), August 09–13, 2026, Jeju Island, Republic of Korea CCS: Computing methodologies Machine learning CCS: Information systems Data mining by Xixun Lin, Zhiheng Zhou, Zhengyin Zhang, Yancheng Chen, Shuai Zhang, Ge Zhang, Shichao Zhu, Lixin Zou, Chuan Zhou, Peng Zhang, Shirui Pan, Yanan Cao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-30786.txt.
Description
Lin and colleagues propose AbstainGNN, a theory-driven design for graph classification with abstention, letting networks reject unsure guesses. Old networks are forced to predict even under high doubt, producing unreliable decisions. AbstainGNN models both the predictive function and the refusal function, using graph structural information. Unlike heuristic methods, the authors characterize the trade-off between errors and rejection costs, deriving a unified learning objective. Training runs in two stages: predictive warm-start, then refusal calibration. On five benchmark datasets, AbstainGNN beats existing refusal methods at the same rejection rates.
Catalog record
| CREATOR | Xixun Lin, Zhiheng Zhou, Zhengyin Zhang, Yancheng Chen, Shuai Zhang, Ge Zhang, Shichao Zhu, Lixin Zou, Chuan Zhou, Peng Zhang, Shirui Pan, Yanan Cao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 65724 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.30786