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 2107.05978cs.LGcs.AIcs.CY
收藏

DIVINE:数据可视化和模型精化的多种影响训练点

DIVINE: Diverse Influential Training Points for Data Visualization and Model Refinement

University of Cambridge(剑桥大学);Amazon AWS AI(亚马逊云科技人工智能部门);The Alan Turing Institute(艾伦·图灵研究所)

Umang Bhatt, Isabel Chien, Muhammad Bilal Zafar, Adrian Weller

摘要:

随着机器学习(ML)模型的复杂性增加,导致预测可解释性的缺乏,已经开发了几种方法来根据对模型影响最大的训练数据点来解释模型的行为。然而,这些方法倾向于将离群值标记为高度影响力的点,限制了从业者可以从不代表训练数据的点中得出的见解。在这项工作中,我们采取了一个步骤,寻找有影响力的训练点,也代表了训练数据。我们首先回顾为训练点分配重要性分数的方法。给定重要性分数,我们提出了一种方法来选择一组DIVerse INfluEntial(DIVINE)训练点作为模型行为的有用解释。由于从业者可能不仅对发现对模型准确性有影响的数据点感兴趣,而且对其他重要指标也感兴趣,因此我们展示了如何基于组公平性来评估训练数据点。我们的方法可以识别不公平的诱导训练点,可以将其删除以改善公平结果。我们的定量实验和用户研究表明,可视化DIVINE点可以帮助从业者更好地理解和解释模型行为。

英文摘要:

As the complexity of machine learning (ML) models increases, resulting in a lack of prediction explainability, several methods have been developed to explain a model's behavior in terms of the training data points that most influence the model. However, these methods tend to mark outliers as highly influential points, limiting the insights that practitioners can draw from points that are not representative of the training data. In this work, we take a step towards finding influential training points that also represent the training data well. We first review methods for assigning importance scores to training points. Given importance scores, we propose a method to select a set of DIVerse INfluEntial (DIVINE) training points as a useful explanation of model behavior. As practitioners might not only be interested in finding data points influential with respect to model accuracy, but also with respect to other important metrics, we show how to evaluate training data points on the basis of group fairness. Our method can identify unfairness-inducing training points, which can be removed to improve fairness outcomes. Our quantitative experiments and user studies show that visualizing DIVINE points helps practitioners understand and explain model behavior better than earlier approaches.

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。

↑