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 1804.00657cs.CV
收藏

从不变性到图像变换的置信度

Confidence from Invariance to Image Transformations

Weizmann Institute of Science(魏茨曼科学研究所);TTIC(丰田技术研究所芝加哥分校)

Yuval Bahat, Gregory Shakhnarovich

摘要:

我们开发了一种自动检测预训练的视觉分类器的分类错误的技术。我们的方法是不可知的分类器的形式,只需要访问一组输入的分类器响应。我们训练一个参数化的二元分类器(错误/正确)的表示来自一组分类器的响应产生的多个副本相同的输入,每个受到不同的自然图像变换。因此,我们建立了一个衡量的信心,在分类器的决策,通过分析其决策的不变性下各种转换。在多个数据集(STL-10,CIFAR-100,ImageNet)和分类器的实验中,我们展示了错误检测任务的最新技术水平。此外,我们将我们的技术应用于新颖性检测场景,在那里我们还展示了最先进的结果。

英文摘要:

We develop a technique for automatically detecting the classification errors of a pre-trained visual classifier. Our method is agnostic to the form of the classifier, requiring access only to classifier responses to a set of inputs. We train a parametric binary classifier (error/correct) on a representation derived from a set of classifier responses generated from multiple copies of the same input, each subject to a different natural image transformation. Thus, we establish a measure of confidence in classifier's decision by analyzing the invariance of its decision under various transformations. In experiments with multiple data sets (STL-10,CIFAR-100,ImageNet) and classifiers, we demonstrate new state of the art for the error detection task. In addition, we apply our technique to novelty detection scenarios, where we also demonstrate state of the art results.

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。

↑