Despite the super-human accuracy of recent deep models in NLP tasks, their robustness is reportedly limited due to their reliance on spurious patterns. We thus aim to leverage contrastive learning and counterfactual augmentation for robustness. For augmentation, existing work either requires humans to add counterfactuals to the dataset or machines to automatically matches near-counterfactuals already in the dataset. Unlike existing augmentation is affected by spurious correlations, ours, by synthesizing "a set"of counterfactuals, and making a collective decision on the distribution of predictions on this set, can robustly supervise the causality of each term. Our empirical results show that our approach, by collective decisions, is less sensitive to task model bias of attribution-based synthesis, and thus achieves significant improvements, in diverse dimensions: 1) counterfactual robustness, 2) cross-domain generalization, and 3) generalization from scarce data.
|Title of host publication||AAAI-22 Technical Tracks 10|
|Publisher||Association for the Advancement of Artificial Intelligence|
|Number of pages||9|
|ISBN (Electronic)||1577358767, 9781577358763|
|Publication status||Published - 2022 Jun 30|
|Event||36th AAAI Conference on Artificial Intelligence, AAAI 2022 - Virtual, Online|
Duration: 2022 Feb 22 → 2022 Mar 1
|Name||Proceedings of the 36th AAAI Conference on Artificial Intelligence, AAAI 2022|
|Conference||36th AAAI Conference on Artificial Intelligence, AAAI 2022|
|Period||22/2/22 → 22/3/1|
Bibliographical noteFunding Information:
This work was supported by SNU-NAVER Hyperscale AI Center and IITP granted by the Korea government (MSIT) [NO.2021-0-0268, AI Innovation Hub (SNU)] and SNU AI Graduate School Program [No.2021-0-01343]. Hwang is a corresponding author.
Copyright © 2022, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.
All Science Journal Classification (ASJC) codes
- Artificial Intelligence