Description

In this thesis, we establish the theoretical basis of some dimensional reduction algorithms like the GPLVM and their application to the reproduction of a time series of observable data with the GPDM and its generalization with control CGPDM. Finally, we are going to introduce a new more time efficient model MoCGPDM applying a mixture of experts. And the final section will consist in fine-tuning the model and compare it to the previous model.

UPCommons: https://upcommons.upc.edu/handle/2117/357951

The work is under the scope of the following projects:

• MdM: Unit of Excellence María de Maeztu (web)
• CLOTHILDE: Cloth manipulation learning from demonstration (web)