Manifold learning

Datum konání: 31.03.2017
Přednášející: Jindřich Soukup
Odpovědná osoba: Soukup

Manifold learning methods search low-dimensional representation of the data. Compared to PCA, manifold learning method preserve local/geodetic distances (not global ones) which helps them to better adapt to a wider range of data. The presentation will include an overview of methods and their application in computer vision.