arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2302.01381cs.LGcs.CV
收藏

针对具有不同训练数据的模型对自然分布偏移的有效鲁棒性

Effective Robustness against Natural Distribution Shifts for Models with Different Training Data

UCLA(加州大学洛杉矶分校);Google Research(谷歌研究院);University of Washington(华盛顿大学);Google(谷歌);OpenAI;UCSB(加州大学圣塔芭芭拉分校)

Zhouxing Shi, Nicholas Carlini, Ananth Balashankar, Ludwig Schmidt, Cho-Jui Hsieh, Alex Beutel, Yao Qin

AI总结:

本文针对不同训练数据模型的有效鲁棒性评估问题,提出一种基于多个分布内测试集的新评估指标,揭示了零样本 CLIP 类模型在单一 ImageNet 测试下的有效鲁棒性增益会被削弱。

摘要:

“有效鲁棒性”衡量的是超出根据分布内(ID)性能所能预测的额外分布外(OOD)鲁棒性。现有的有效鲁棒性评估通常使用单一测试集(如 ImageNet)来评估 ID 准确率。当评估在不同数据分布上训练的模型时,这会产生问题,例如,比较在 ImageNet 上训练的模型与在 LAION 上训练的零样本语言-图像预训练模型。在本文中,我们提出了一种新的评估指标,用于评估和比较在不同数据上训练的模型的有效鲁棒性。为此,我们控制了多个 ID 测试集上的准确率,这些测试集覆盖了所有被评估模型的训练分布。当存在具有不同训练数据的模型时,我们的新评估指标能够更好地估计有效鲁棒性。它还可以解释先前工作中表现出的零样本 CLIP 类模型令人惊讶的有效鲁棒性提升,这些工作仅使用 ImageNet 作为唯一的 ID 测试集,而在我们的新评估下,这种提升有所减弱。更多工件(包括交互式可视化)提供于 https://shizhouxing.github.io/effective-robustness。

英文摘要:

"Effective robustness" measures the extra out-of-distribution (OOD) robustness beyond what can be predicted from the in-distribution (ID) performance. Existing effective robustness evaluations typically use a single test set such as ImageNet to evaluate the ID accuracy. This becomes problematic when evaluating models trained on different data distributions, e.g., comparing models trained on ImageNet vs. zero-shot language-image pre-trained models trained on LAION. In this paper, we propose a new evaluation metric to evaluate and compare the effective robustness of models trained on different data. To do this, we control for the accuracy on multiple ID test sets that cover the training distributions for all the evaluated models. Our new evaluation metric provides a better estimate of effective robustness when there are models with different training data. It may also explain the surprising effective robustness gains of zero-shot CLIP-like models exhibited in prior works that used ImageNet as the only ID test set, while the gains diminish under our new evaluation. Additional artifacts including interactive visualizations are provided at https://shizhouxing.github.io/effective-robustness.

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。

↑