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 1908.04626cs.CL
收藏

注意力不是解释

Attention is not not Explanation

Georgia Institute of Technology(佐治亚理工学院)

Sarah Wiegreffe, Yuval Pinter

摘要:

注意力机制在自然语言处理系统中起着核心作用,特别是在递归神经网络(RNN)模型中。最近,人们越来越感兴趣的是,这些模块提供的中间表示是否可以用来解释模型预测的推理,从而获得关于模型决策过程的见解。最近的一篇论文声称“注意力不是解释”(Jain和Wallace,2019)。我们对这项工作的许多假设提出了挑战,认为这种说法取决于人们对解释的定义,并且测试它需要考虑模型的所有元素,使用严格的实验设计。我们提出了四种替代测试来确定何时/是否可以使用注意力作为解释:简单的统一权重基线;基于多个随机种子运行的方差校准;使用预训练模型的冻结权重的诊断框架;以及端到端的对抗性注意力训练协议。每个都允许对RNN模型中的注意力机制进行有意义的解释。我们发现,即使可以找到可靠的对抗分布,他们不执行简单的诊断,表明以前的工作并没有反驳注意力机制的可解释性的有用性。

英文摘要:

Attention mechanisms play a central role in NLP systems, especially within recurrent neural network (RNN) models. Recently, there has been increasing interest in whether or not the intermediate representations offered by these modules may be used to explain the reasoning for a model's prediction, and consequently reach insights regarding the model's decision-making process. A recent paper claims that `Attention is not Explanation' (Jain and Wallace, 2019). We challenge many of the assumptions underlying this work, arguing that such a claim depends on one's definition of explanation, and that testing it needs to take into account all elements of the model, using a rigorous experimental design. We propose four alternative tests to determine when/whether attention can be used as explanation: a simple uniform-weights baseline; a variance calibration based on multiple random seed runs; a diagnostic framework using frozen weights from pretrained models; and an end-to-end adversarial attention training protocol. Each allows for meaningful interpretation of attention mechanisms in RNN models. We show that even when reliable adversarial distributions can be found, they don't perform well on the simple diagnostic, indicating that prior work does not disprove the usefulness of attention mechanisms for explainability.

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。

↑