Multi-level statistical analysis of high-frequency spatio-temporal process data (C03)

高频高维过程数据通常分几个步骤进行分析。由于存储容量有限，需要实时识别和选择感兴趣的数据进行进一步分析。通过对时空数据中以瞬变模式出现的天体物理粒子的检测和分类这一极具挑战性的例子，我们希望构建用于对这类数据进行简化、分类和建模的健壮的、可推广的技术，以及适当结合这些分析步骤的策略。

High-frequency high-dimensional process data are typically analyzed in several steps. Interesting data need to be identified and selected in real time for further analysis because of limited storage capacities. By means of the challenging example of detection and classification of astrophysical particles appearing as transient patterns in spatio-temporal data, we want to construct robust generalizable techniques for reduction, classification, and modeling of such data, as well as strategies for the suitable combination of these analysis steps.