Quantization and clustering on Riemannian manifolds with an application to air traffic analysis (Q2274972)

In various applications in computer vision, contour analysis and other fields, signals belong not to Euclidean vector space, but rather to a Riemannian manifold. An \textit{online} quantization algorithm for big data is proposed in sense that at each step it computes as many distances as the number of desired clusters, which is a small number. It continues \textit{G. Pagès}' studies [ESAIM, Proc. Surv. 48, 29--79 (2015; Zbl 1338.60114)] developed for Euclidean vector space. The efficiency of the proposed method is illustrated on French air traffic analysis.