Grant PID2021-123090NB-I00 funded by: 
This project promotes the research and development of novel unsupervised and supervised learning techniques that can effectively blend the model-based and data-driven paradigms. We aim to use the model-based information, hypotheses, and algorithmic structure to effectively regularize the existing learning techniques and overcome some of their current limitations, such as the lack of interpretability and identifiability in their solutions, the overparameterization, and the need for large training sets. We propose to apply the resulting techniques to challenging problems, such as improving the EEG processing techniques for brain-computer interfaces, the development of algorithmic tools for the integration of emotion recognition and virtual reality, as well as the research on state-of-the-art methods and distributed learning techniques for the identification of behavioral models and the predistortion of power amplifiers. In summary, the general objectives of the project are as follows.
- Propose learning techniques that can enable the effective integration of model-based and data-driven paradigms. To pursue the convergence to disentangled representations, as well as to research conditions that allow for the identifiability and interpretability of the latent features of the observations.
- To develop novel signal processing and machine learning techniques for the processing of EEG signals with applications in brain-computer interfaces, and the exploration of the interplay between EEG-based emotion/attention recognition and virtual reality.
- To benchmark the performance of novel behavioral models for communication systems and computationally efficient linearization schemes, including those that put into practice distributed and federated learning techniques, as well as other centralized machine learning or Bayesian approaches.
- As a complement to the above objectives, we also consider the opportunity of the project to maintain and consolidate the international expertise of the team in each of the proposed research areas, as well as for the training of the requested students within the framework of our doctoral program and through research stays in the prestigious institutions of our collaborators.