{"id":"arxiv/2605.01624","url":"https://arxiv.org/abs/2605.01624","tags":["math.AT","cs.LG","stat.AP","stat.ML"],"score":0,"title":"Persistent Homology of Time Series through Complex Networks","domain":"math.AT","source":"arxiv","authors":["İsmail Güzel"],"reviews":0,"subject":"math.AT","summary":"We present a unified pipeline for univariate time series classification via complex networks and persistent homology. A time series is mapped to a graph through one of five constructions across three families (visibility (natural and horizontal visibility graphs), transition, and proximity) and the graph is converted to a dissimilarity matrix from which a Vietoris-Rips filtration yields persistence diagrams. These diagrams are vectorized into fixed-length features through persistence landscapes and topological summary statistics. By standardizing the downstream processing, differences in…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-05-02","reactions":{}}