arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2107.06997cs.LGcs.AIcs.SE
收藏

DeepHyperion:通过光照搜索探索基于深度学习的系统的特征空间

DeepHyperion: Exploring the Feature Space of Deep Learning-Based Systems through Illumination Search

Università della Svizzera Italiana(瑞士意大利语大学);University of Passau(帕绍大学)

Tahereh Zohdinasab, Vincenzo Riccio, Alessio Gambi, Paolo Tonella

摘要:

深度学习(DL)已成功应用于广泛的应用领域,包括安全关键领域。最近在文献中提出了几种DL测试方法,但没有一种旨在评估所生成的输入的不同可解释特征如何影响系统的行为。在本文中,我们求助于照明搜索来找到性能最高的测试用例(即,行为不当和最接近行为不当),分布在表示系统的特征空间的图的单元上。我们介绍了一种方法,指导我们的方法的用户在识别和量化的特征空间的维度为一个给定的域的任务。我们开发了深度学习系统的基于搜索的工具Deepcup,它可以说明,即,通过为开发人员提供可解释的特征图,在特征图中,自动生成的输入与有关暴露行为的信息一起放置,从而探索特征空间。

英文摘要:

Deep Learning (DL) has been successfully applied to a wide range of application domains, including safety-critical ones. Several DL testing approaches have been recently proposed in the literature but none of them aims to assess how different interpretable features of the generated inputs affect the system's behaviour. In this paper, we resort to Illumination Search to find the highest-performing test cases (i.e., misbehaving and closest to misbehaving), spread across the cells of a map representing the feature space of the system. We introduce a methodology that guides the users of our approach in the tasks of identifying and quantifying the dimensions of the feature space for a given domain. We developed DeepHyperion, a search-based tool for DL systems that illuminates, i.e., explores at large, the feature space, by providing developers with an interpretable feature map where automatically generated inputs are placed along with information about the exposed behaviours.

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。

↑