
Topologically-Constrained Latent Variable Models
Published on 2008-07-2911407 Views
In dimensionality reduction approaches, the data are typically embedded in a Euclidean latent space. However for some data sets this is inappropriate. For example, in human motion data we expect laten