标题： 将机器学习应用于天体物理学中的目录匹配 显示英文标题

标题： Applying Machine Learning to Catalogue Matching in Astrophysics

摘要： 我们展示了将自动化机器学习技术应用于天文学中不同目标目录匹配问题的结果。在这项研究中，我们使用了两个部分匹配的目录，其中一个目录具有较大的位置不确定性。我们使用的两个目录来自HI Parkes全天巡天（HIPASS）和SuperCOSMOS光学巡天。先前的工作已经将44%（1887个目标）的HIPASS与SuperCOSMOS目录进行了匹配。然后应用监督学习算法构建了目录中已匹配部分的模型。模型验证显示我们获得了良好的分类性能（99.12%正确）。应用该模型到未匹配的部分目录中，找到了1209个新的匹配。这使目录的匹配对象数量从1887个增加到3096个。这些程序的结合产生了一个72%匹配的目录。 显示英文摘要

摘要： We present the results of applying automated machine learning techniques to the problem of matching different object catalogues in astrophysics. In this study we take two partially matched catalogues where one of the two catalogues has a large positional uncertainty. The two catalogues we used here were taken from the HI Parkes All Sky Survey (HIPASS), and SuperCOSMOS optical survey. Previous work had matched 44% (1887 objects) of HIPASS to the SuperCOSMOS catalogue. A supervised learning algorithm was then applied to construct a model of the matched portion of our catalogue. Validation of the model shows that we achieved a good classification performance (99.12% correct). Applying this model, to the unmatched portion of the catalogue found 1209 new matches. This increases the catalogue size from 1887 matched objects to 3096. The combination of these procedures yields a catalogue that is 72% matched.