Descripció del projecte

Decode non-invasive electroencephalography (EEG) signals and use them as signals to re-train LLMs using Reinforcement Learning, Decision Transformers, or State-Space models. Also research multi-objective strategies that can address theoretical and algorithmic impasses that single-objective methods cannot, such as:(1) a priori and uncertain decisions made about desired trade-offs between objectives, (2) not considering multiple objectives or collapsing the objectives to single scalar rewards, and (3) limited ability to discover multiple policies, so as to rapidly adapt to changing preferences, and apply bi-directional decision making. Conduct formal analyses of neurophysiological correlates of behavioural indicators, to drive our understanding of how relevant cognitive phenomena (i.e., decision making, perception, etc.) occur in human neurophysiology. This analysis will be based on a cross-modal neuroimaging approach that fuses simultaneously-acquired EEG/fNIRS data (with other sensory channels), via multivariate single-trial analyses and computational modelling techniques.



MÉS INFORMACIÓ

Si t’interessa l’oferta, omple el pdf amb les teves dades i envia´l a doctorats.industrials.recerca@gencat.cat