Freya Behrens

Freya Behrens is a Researcher at the Statistical Physics of Computation Laboratory (SB) within the École Polytechnique Fédérale de Lausanne (EPFL). Her work bridges statistical physics and computational systems, focusing on topics like neural networks, cellular automata, and AI applications in education. She investigates phase transitions in attention mechanisms and the theoretical underpinnings of deep learning. Recent publications explore AI's impact on higher education and the interplay between Hessian matrices and decision boundaries.

Her research integrates interdisciplinary methods from physics and computer science, addressing questions such as how memorization occurs in neural networks and the dynamics of spin glasses. Freya’s contributions span theoretical frameworks (e.g., dynamical cavity methods) and applied AI systems (e.g., knowledge graph exploration).

While no specific grants or awards are listed, her work demonstrates engagement with cutting-edge challenges in machine learning and computational modeling. She is affiliated with EPFL’s School of Basic Sciences, reflecting her foundational research in physics and mathematics-driven computation.