Here you will find some results of the embedding of high-dimensional data points in a low-dimensional Euclidean space using diffusion maps. We focus on the analysis of multivariate recordings of EEG signals in different contexts.
Diffusion maps are part of what researchers call manifold learning theory and are often used for visualization purposes. However, it can also provide very interesting information about the intrinsic geometry of a low-dimensional manifold where the data points are assumed to live in. In fact, one can even do semi-supervised learning using the information gathered by the diffusion maps.
These are two PDF files with presentations I gave at the GIPSA-lab in March/2017
If after reading this page you still have any questions or just want to discuss about something related to diffusion maps (or not), please feel free to send an e-mail to: pedro-luiz.coelho-rodrigues at gipsa-lab dot fr