CLiPS Colloquium - Helena Balabin: Natural Language Processing in the Human Brain
Abstract
The intersection of Natural Language Processing and neurolinguistics holds potential for advancing language understanding in both artificial intelligence and neuroscience. This talk provides an overview of ongoing work at this intersection, comprising both clinical and fundamental research. First, the application of Dutch language models for the prediction of early Alzheimer’s disease from connected speech is discussed. Next, emerging shared patterns between language model embeddings and structural magnetic resonance imaging (MRI) data are examined through representational similarity analysis. The second part of the talk is dedicated towards linking foundation models to brain activity patterns derived from functional MRI experiments using neural en- and decoding. This approach allows for both (i) the localization of specified cognitive processes in the brain and (ii) a biologically plausible evaluation of foundation model embeddings.
Bio
Helena Balabin is an interdisciplinary PhD student at the departments of Neurosciences and Computer Science at KU Leuven, jointly supervised by Prof. Rik Vandenberghe, Prof. Patrick Dupont and Prof. Marie-Francine (Sien) Moens. Her research lies at the intersection of Natural Language Processing and neurolinguistics, ranging from fundamental research to clinical applications. Prior to starting her PhD in Belgium, she completed her Master’s in Computer Science at the Bonn-Rhein-Sieg University of Applied Sciences in Sankt Augustin, Germany. During that time, she has also worked as a student research assistant at the Bioinformatics department of the Fraunhofer Institute for Algorithms and Scientific Computing.
Program
15:00 - Presentation
15:45 - Coffee break
16:00 - Discussion
Date
14th of March 2025
Location
S.KS 203
Registration
To register for the colloquium, please fill in this form.