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 2206.01367cs.LGcs.CR
收藏

对抗性遗忘:沿着对抗性方向减少信心

Adversarial Unlearning: Reducing Confidence Along Adversarial Directions

Carnegie Mellon University(卡内基梅隆大学);UC Berkeley(加州大学伯克利分校)

Amrith Setlur, Benjamin Eysenbach, Virginia Smith, Sergey Levine

摘要:

用最大似然目标训练的监督学习方法经常在训练数据上过拟合。大多数防止过拟合的正则化器看起来是为了增加对其他示例的信心(例如,数据增强、对抗训练),或在训练数据上减少它(例如,标签平滑)。在这项工作中,我们提出了一个互补的正则化策略,降低了对自我生成的例子的信心。我们称之为RCAD(沿对抗方向降低置信度)的方法旨在降低沿对抗方向分布的样本的置信度,以增加训练损失。与对抗训练相比,RCAD并不试图对模型进行鲁棒性处理以输出原始标签,而是将其正则化,以降低使用比传统对抗训练更大的扰动生成的点的置信度。RCAD可以通过几行代码轻松集成到培训管道中。尽管它很简单,但我们发现在许多分类基准上,RCAD可以添加到现有技术中(例如,标签平滑,MixUp训练),以将测试精度的绝对值提高1-3%,在低数据状态下有更显著的增益。我们还提供了一个理论分析,有助于在简化的设置中解释这些好处,表明RCAD可以证明帮助模型忘记训练数据中的虚假特征。

英文摘要:

Supervised learning methods trained with maximum likelihood objectives often overfit on training data. Most regularizers that prevent overfitting look to increase confidence on additional examples (e.g., data augmentation, adversarial training), or reduce it on training data (e.g., label smoothing). In this work we propose a complementary regularization strategy that reduces confidence on self-generated examples. The method, which we call RCAD (Reducing Confidence along Adversarial Directions), aims to reduce confidence on out-of-distribution examples lying along directions adversarially chosen to increase training loss. In contrast to adversarial training, RCAD does not try to robustify the model to output the original label, but rather regularizes it to have reduced confidence on points generated using much larger perturbations than in conventional adversarial training. RCAD can be easily integrated into training pipelines with a few lines of code. Despite its simplicity, we find on many classification benchmarks that RCAD can be added to existing techniques (e.g., label smoothing, MixUp training) to increase test accuracy by 1-3% in absolute value, with more significant gains in the low data regime. We also provide a theoretical analysis that helps to explain these benefits in simplified settings, showing that RCAD can provably help the model unlearn spurious features in the training data.

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。

↑