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 1908.08142cs.LGcs.CVstat.ML
收藏

监督分类任务的可迁移性和难度

Transferability and Hardness of Supervised Classification Tasks

VinAI Research(VinAI研究院);Amazon Web Services(亚马逊云科技);Facebook AI(脸书人工智能)

Anh T. Tran, Cuong V. Nguyen, Tal Hassner

摘要:

我们提出了一种新的方法来估计监督分类任务的难度和可转移性。与以前的工作不同,我们的方法是解决方案不可知的,不需要或假设训练模型。相反,我们使用信息论方法来估计这些值:将训练标签视为随机变量并探索其统计数据。当从源任务转移到目标任务时,我们考虑两个这样的变量之间的条件熵(即,两个任务的标签分配)。我们的分析和经验表明,这个值是有关的损失转移模型。我们进一步展示了如何使用这个值来估计任务难度。我们在三个大规模数据集上广泛测试了我们的主张- CelebA(40个任务),具有属性的动物2(85个任务)和Caltech-UCSD鸟类200(312个任务)-共代表437个分类任务。我们提供的结果表明,我们的硬度和可转移性的估计是强烈相关的经验硬度和可转移性。作为一个案例研究,我们将一个学习的人脸识别模型转移到CelebA属性分类任务中,显示了估计高度可转移的任务的最新准确性。

英文摘要:

We propose a novel approach for estimating the difficulty and transferability of supervised classification tasks. Unlike previous work, our approach is solution agnostic and does not require or assume trained models. Instead, we estimate these values using an information theoretic approach: treating training labels as random variables and exploring their statistics. When transferring from a source to a target task, we consider the conditional entropy between two such variables (i.e., label assignments of the two tasks). We show analytically and empirically that this value is related to the loss of the transferred model. We further show how to use this value to estimate task hardness. We test our claims extensively on three large scale data sets -- CelebA (40 tasks), Animals with Attributes 2 (85 tasks), and Caltech-UCSD Birds 200 (312 tasks) -- together representing 437 classification tasks. We provide results showing that our hardness and transferability estimates are strongly correlated with empirical hardness and transferability. As a case study, we transfer a learned face recognition model to CelebA attribute classification tasks, showing state of the art accuracy for tasks estimated to be highly transferable.

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。

↑