Applying machine learning techniques to detect and classify faint tidal features with samples from upcoming large-scale surveys.

微弱的潮汐特征是星系形成模型和Lambda-CDM宇宙学模型的关键预测。现代即将到来的全天空调查，如LSST和欧几里得应该能够检测到数百万个这样的特征;然而，人类分类将无法扩展到这些数量。为此，该项目将使用自动化技术，从这些调查中确定潮汐特征并对其进行分类，减少人工检查的需要，从而增加统计样本的规模和发现机会。特别是，该项目将利用机器学习。机器学习在天文学中的使用激增，部分原因是天文数据集的大量增加和计算硬件的进步。以前的工作已经显示了检测潮汐特征的可能性，但样本量需要更大，才能在分类和分割方面取得重大进展。来自即将到来的大规模调查的数据量将改变这一点，从而使人们能够更深入地了解星系的形成和过程。

Faint tidal features are a key prediction of galaxy formation models and the Lambda-CDM cosmological model. Modern upcoming all-sky surveys such as LSST and Euclid should be capable of detecting millions of these features; however, human classification will not scale to these volumes. To that end, this project will use automated techniques to identify and classify tidal features from these surveys, reducing the need for human inspection and thus increasing the size of statistical samples and opportunities for discovery. In particular, the project will make use of machine learning. Machine learning has seen an upsurge in use across astronomy due in part to the vast increase in astronomical datasets and advancements in computational hardware. Previous work has shown the possibility of detecting tidal features, but the sample sizes need to be bigger to make significant advancements towards classification and segmentation. The amount of data from upcoming large-scale surveys will change that and in turn allow for deeper insight into galaxy formation and processes.