Dimensionality Estimation

Quadric Hypersurface Intersection for Manifold Learning in Feature Space
A novel approach to use the so-called “Manifold Hypothesis” in order to solve the outlier detection problem by quadrics on top of FaceID embeddings.
Fedor Pavutnitskiy, Sergei O. Ivanov, Evgeniy Abramov, Viacheslav Borovitskiy, Artem Klochkov, Viktor Vyalov, Anatolii Zaikovskii, Aleksandr Petiushko Александр Петюшко
PDF Cite Code Video Source Document arXiv
Quadric Hypersurface Intersection for Manifold Learning in Feature Space
Manifold Hypothesis in Data Analysis: Double Geometrically-Probabilistic Approach to Manifold Dimension Estimation
Elegant ways to check the so-called “Manifold Hypothesis” for FaceID using geoemtric (Ninkiwsky-based) and probabilistic approaches.
Alexander Ivanov, Gleb Nosovskiy, Alexey Chekunov, Denis Fedoseev, Vladislav Kibkalo, Mikhail Nikulin, Fedor Popelenskiy, Stepan Komkov, Ivan Mazurenko, Aleksandr Petiushko Александр Петюшко
PDF Cite DOI arXiv
Manifold Hypothesis in Data Analysis: Double Geometrically-Probabilistic Approach to Manifold Dimension Estimation