arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.14252cs.LGcs.CL
收藏

对机器生成文本黑盒检测器解释方法的评估

An Evaluation of Explanation Methods for Black-Box Detectors of Machine-Generated Text

Faculty of Computer Science, University of Vienna(维也纳大学计算机科学系);UniVie Doctoral School Computer Science, University of Vienna(维也纳大学计算机科学博士学院);Faculty of Philological and Cultural Studies, University of Vienna(维也纳大学文学与文化研究系)

Loris Schoenegger, Yuxi Xia, Benjamin Roth

AI总结:

本文评估了三种解释方法在机器生成文本检测器中的表现,发现SHAP在忠实度和稳定性方面最佳,而LIME在用户表现上最差。

摘要:

随着区分语言模型生成文本与人工写作文本的难度增加,已经开发出检测机器生成文本(MGT)的检测器。然而,在许多情况下,黑盒预测本身并不足够,了解检测器做出预测的依据同样重要。能够估计特征重要性的解释方法有望表明分类器在预测中使用了输入的哪些部分。然而,这些方法通常使用简单的分类器和对人类来说直观的任务进行评估。为了评估其在更广泛情境下的适用性,本研究进行了首次系统评估解释质量对于MGT检测器的适用性。通过五个自动化实验评估了忠实度和稳定性两个维度,并通过用户研究评估了实用性。我们使用了由ChatGPT生成和人工写作的文档数据集,并将三种现有语言模型基检测器的预测与相应的SHAP、LIME和Anchor解释进行配对。我们发现SHAP在忠实度、稳定性以及帮助用户预测检测器行为方面表现最佳。相比之下,LIME被认为对用户最有用,但在预测检测器行为的用户表现方面得分最差。

英文摘要:

The increasing difficulty to distinguish language-model-generated from human-written text has led to the development of detectors of machine-generated text (MGT). However, in many contexts, a black-box prediction is not sufficient, it is equally important to know on what grounds a detector made that prediction. Explanation methods that estimate feature importance promise to provide indications of which parts of an input are used by classifiers for prediction. However, these are typically evaluated with simple classifiers and tasks that are intuitive to humans. To assess their suitability beyond these contexts, this study conducts the first systematic evaluation of explanation quality for detectors of MGT. The dimensions of faithfulness and stability are evaluated with five automated experiments, and usefulness is assessed in a user study. We use a dataset of ChatGPT-generated and human-written documents, and pair predictions of three existing language-model-based detectors with the corresponding SHAP, LIME, and Anchor explanations. We find that SHAP performs best in terms of faithfulness, stability, and in helping users to predict the detector's behavior. In contrast, LIME, perceived as most useful by users, scores the worst in terms of user performance at predicting detector behavior.

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。

↑