Exploring local interpretability in dimensionality reduction: Analysis and use cases

Nikolaos Mylonas, Ioannis Mollas, Nick Bassiliades, Grigorios Tsoumakas

Abstract: Highlights•Non-linear dimensionality reduction is necessary in several use cases.•Interpretability is vital for most of those use cases.•LXDR agnostically interprets any dimensionality reduction technique.•LXDR is applicable to various domains and use cases.