Hangjin Jiang, Yuzhou Li, Zhaoxing Gao
This paper introduces a distributed framework for dimensionality reduction in high-dimensional matrix-variate time series, improving computational efficiency and information utilization through a split-and-conquer approach using tensor PCA.
The researchers have developed a new method to handle large and complex datasets that come in the form of time series arranged in matrices. By breaking these large datasets into smaller parts and processing them separately on different computers, they can efficiently reduce the data's complexity without losing important information. This method not only speeds up computation but also maintains the data's inherent structure better than previous methods. The approach is tested through simulations and real-world data, showing it can accurately predict and analyze trends over time.
