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 1909.07913cs.CLcs.LG
收藏

学习使用基于注意力的推理进行欺骗

Learning to Deceive with Attention-Based Explanations

Carnegie Mellon University(卡内基梅隆大学);Twitter(推特)

Danish Pruthi, Mansi Gupta, Bhuwan Dhingra, Graham Neubig, Zachary C. Lipton

摘要:

注意力机制是应用于自然语言处理的神经结构中普遍存在的组件。除了在预测准确性方面获得收益外,注意力权重通常被认为具有可解释性,据称对于为从业者提供见解以及解释模型为何向利益相关者做出决策都很有用。我们呼吁后者使用的注意力机制的问题,展示了一个简单的方法,训练模型,以产生欺骗性的注意力面具。我们的方法减少了分配给指定的不允许令牌的总权重,即使模型可以被证明仍然依赖于这些特征来驱动预测。在多个模型和任务中,我们的方法可以操纵注意力权重,同时在准确性方面付出的代价非常小。通过一项人类研究,我们表明,我们操纵的基于注意力的解释欺骗人们认为,对性别少数群体有偏见的模型的预测不依赖于性别。因此,我们的研究结果怀疑注意力的可靠性作为审计算法的工具,在公平性和问责制的背景下。

英文摘要:

Attention mechanisms are ubiquitous components in neural architectures applied to natural language processing. In addition to yielding gains in predictive accuracy, attention weights are often claimed to confer interpretability, purportedly useful both for providing insights to practitioners and for explaining why a model makes its decisions to stakeholders. We call the latter use of attention mechanisms into question by demonstrating a simple method for training models to produce deceptive attention masks. Our method diminishes the total weight assigned to designated impermissible tokens, even when the models can be shown to nevertheless rely on these features to drive predictions. Across multiple models and tasks, our approach manipulates attention weights while paying surprisingly little cost in accuracy. Through a human study, we show that our manipulated attention-based explanations deceive people into thinking that predictions from a model biased against gender minorities do not rely on the gender. Consequently, our results cast doubt on attention's reliability as a tool for auditing algorithms in the context of fairness and accountability.

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。

↑