DIKU Bits: Mapping Brains with Language Models

Speaker

Antonia Karamolegkou, PhD fellow in the Natural Language Processing (NLP) section at the Department of Computer Science, University of Copenhagen

Title

Mapping Brains with Language Models

Abstract

Over the years, many researchers have made the same observation: Human brain and language model activations exhibit structural similarities, assuming a linear mapping between features extracted from neural recordings (fMRI, MEG) and computational language models. How much evidence has been accumulated, what exactly drives this alignment, and what are we to understand from it? In this talk, we will answer these questions by focusing both on studies that use models to explain brain activity during language processing, and studies that use neuroimaging data to enhance language models.

Which courses do you teach?

Introduction to Data Science, Natural Language Processing (NLP)

Which technology/research/projects/startup are you excited to see the evolution of?

The Future of Human-Machine Interaction, Transparent and Explainable AI, Blackbox (brain) vs. Blackbox (neural network)