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 2206.03575cs.LG
收藏

线性回归中数据偏差稳健性的证明

Certifying Data-Bias Robustness in Linear Regression

University of Wisconsin–Madison(威斯康星大学麦迪逊分校)

Anna P. Meyer, Aws Albarghouthi, Loris D'Antoni

摘要:

由于人为错误和社会偏见,数据集通常包含不准确性,这些不准确性可能会影响在此类数据集上训练的模型的结果。我们提出了一种技术,用于证明线性回归模型是否对训练数据集中的标签偏差具有逐点鲁棒性,即,对训练数据集标签的有界扰动是否会导致模型改变测试点的预测。我们展示了如何解决这个问题,正是为个别测试点,并提供了一个近似的,但更可扩展的方法,不需要预先知道的测试点。我们广泛地评估了这两种技术,发现线性模型-回归和分类-通常显示出高水平的偏差鲁棒性。然而,我们也发现了偏差鲁棒性方面的差距,例如某些数据集上某些偏差假设的高水平非鲁棒性。总的来说,我们的方法可以作为何时信任或质疑模型输出的指南。

英文摘要:

Datasets typically contain inaccuracies due to human error and societal biases, and these inaccuracies can affect the outcomes of models trained on such datasets. We present a technique for certifying whether linear regression models are pointwise-robust to label bias in the training dataset, i.e., whether bounded perturbations to the labels of a training dataset result in models that change the prediction of test points. We show how to solve this problem exactly for individual test points, and provide an approximate but more scalable method that does not require advance knowledge of the test point. We extensively evaluate both techniques and find that linear models -- both regression- and classification-based -- often display high levels of bias-robustness. However, we also unearth gaps in bias-robustness, such as high levels of non-robustness for certain bias assumptions on some datasets. Overall, our approach can serve as a guide for when to trust, or question, a model's output.

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。

↑