Stable ranks for data analysis

Hierarchical stabilisation, allows us to define topological invariants for data starting from metrics to compare persistence modules. In this talk I will highlight the variety of metrics that can be constructed in an axiomatic way, via so called Noise Systems. The focus will then be on one invariant obtained through hierarchical stabilisation, the Stable Rank, which the TDA group at KTH has been studying in the last years. In particular I will address the problem of using this invariant on noisy and heterogeneous data. Lastly, I will illustrate the use of stable ranks on real data within a project on microglia morphology description, in collaboration with S. Siegert’s group, K. Hess and L. Kanari.