Enhancing EEG-based brain-computer interfaces through multitask learning

Electroencephalography (EEG) is a method that allows the measurement of brain activity with high-temporal resolution in a continuous and portable way. It could be used in a wide range of use-cases, from medical clinical applications to the development of Brain-Computer Interface (BCI) technologies. When developing BCI systems, it is crucial to identify meaningful attributes from the EEG signals that could be used to train reliable classifiers, and to accurately decode brain signals to control external devices. Multitask Learning (MTL), a machine learning technique, can be employed to address this challenge by simultaneously denoising EEG signals and learning latent representations in order to train a classifier to predict diferent brain states. In light of this, the present research project aims to explore the usage of MTL in the context of an EEG-based BCI system, with the goal of improving the accuracy and reliability of decoding brain signals during motor imagery tasks.