Segmentation of Piecewise ARX Processes by Exploiting Sparsity in Tight-Dimensional Spaces

Segmentation of piecewise Auto-Regressive eXogenous (ARX) processes has been a major challenge in time-series segmentation and change detection. In this paper, for piecewise ARX process segmentation, we exploit hidden sparsity in tight-dimensional representation spaces. More precisely, we strategically design a tight-dimensional linear transformation which reveals sparsity hidden in samples following piecewise ARX processes. Experiments on synthetic and real-world data demonstrate the effectiveness of the proposed method.