arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2211.16093cs.CL
收藏

惩罚大扰动输入下的置信预测不能改善问答中的分布外泛化

Penalizing Confident Predictions on Largely Perturbed Inputs Does Not Improve Out-of-Distribution Generalization in Question Answering

The University of Tokyo(东京大学);National Institute of Informatics(国立信息学研究所)

Kazutoshi Shinoda, Saku Sugawara, Akiko Aizawa

摘要:

问答(QA)模型被证明是不敏感的大扰动输入,也就是说,他们作出正确和自信的预测,即使在很大程度上扰动输入,人类不能正确地得出答案。此外,QA模型无法推广到其他领域和对抗性测试集,而人类则保持了高准确性。基于这些观察,我们假设QA模型不使用人类阅读所需的预期特征,而是依赖于虚假特征,导致缺乏泛化能力。因此,我们试图回答这样一个问题:如果QA模型对各种类型的扰动的过度自信的预测受到惩罚,是否会改善分布外(OOD)的推广?为了防止模型对扰动输入做出自信的预测,我们首先遵循现有的研究,并最大化扰动输入的输出概率的熵。然而,我们发现,训练成对某种扰动类型敏感的QA模型往往对看不见的扰动类型不敏感。因此,我们同时最大化四种扰动类型的熵(即,单词和重复级别的洗牌和删除),以进一步缩小模型和人类之间的差距。与我们的预期相反,虽然模型变得敏感的四种类型的扰动,我们发现,面向对象的推广并没有得到改善。此外,面向对象的推广有时会退化后熵最大化。对很大程度上受到干扰的输入本身进行不自信的预测可能有利于获得人类的信任。然而,我们的负面结果表明,研究人员应该注意熵最大化的副作用。

英文摘要:

Question answering (QA) models are shown to be insensitive to large perturbations to inputs; that is, they make correct and confident predictions even when given largely perturbed inputs from which humans can not correctly derive answers. In addition, QA models fail to generalize to other domains and adversarial test sets, while humans maintain high accuracy. Based on these observations, we assume that QA models do not use intended features necessary for human reading but rely on spurious features, causing the lack of generalization ability. Therefore, we attempt to answer the question: If the overconfident predictions of QA models for various types of perturbations are penalized, will the out-of-distribution (OOD) generalization be improved? To prevent models from making confident predictions on perturbed inputs, we first follow existing studies and maximize the entropy of the output probability for perturbed inputs. However, we find that QA models trained to be sensitive to a certain perturbation type are often insensitive to unseen types of perturbations. Thus, we simultaneously maximize the entropy for the four perturbation types (i.e., word- and sentence-level shuffling and deletion) to further close the gap between models and humans. Contrary to our expectations, although models become sensitive to the four types of perturbations, we find that the OOD generalization is not improved. Moreover, the OOD generalization is sometimes degraded after entropy maximization. Making unconfident predictions on largely perturbed inputs per se may be beneficial to gaining human trust. However, our negative results suggest that researchers should pay attention to the side effect of entropy maximization.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑