Skip to Main content Skip to Navigation
Conference papers

Local intrinsic dimensionality estimators based on concentration of measure

Abstract : Intrinsic dimensionality (ID) is one of the most fundamental characteristics of multi-dimensional data point clouds. Knowing ID is crucial to choose the appropriate machine learning approach as well as to understand its behavior and validate it. ID can be computed globally for the whole data distribution, or estimated locally in a point. In this paper, we introduce new local estimators of ID based on linear separability of multi-dimensional data point clouds, which is one of the manifestations of concentration of measure. We empirically study the properties of these measures and compare them with other recently introduced ID estimators exploiting various other effects of measure concentration. Observed differences in the behaviour of different estimators can be used to anticipate their behaviour in practical applications.
Complete list of metadatas

Cited literature [33 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-02972293
Contributor : Andrei Zinovyev <>
Submitted on : Tuesday, October 20, 2020 - 12:08:49 PM
Last modification on : Thursday, January 14, 2021 - 3:14:13 PM
Long-term archiving on: : Thursday, January 21, 2021 - 6:54:07 PM

File

BacZinovyev_IJCNN2020.pdf
Files produced by the author(s)

Identifiers

Citation

Jonathan Bac, Andrei Zinovyev. Local intrinsic dimensionality estimators based on concentration of measure. 2020 International Joint Conference on Neural Networks (IJCNN), Jul 2020, Glasgow, United Kingdom. pp.1-8, ⟨10.1109/IJCNN48605.2020.9207096⟩. ⟨hal-02972293⟩

Share

Metrics

Record views

32

Files downloads

133