Statistical Biophysics
The Statistical Biophysics group seeks to understand fundamental biological phenomena using tools and concepts from physics, particularly statistical biophysics and stochastic dynamical systems theory.
It pushes beyond existing theoretical frameworks by developing a nonequilibrium physics of living systems and learning machines.
The Statistical Biophysics group currently focuses on dynamics and statistical properties of information processing networks, encompassing biochemical networks in cells, neural circuits in the brain and modern artificial neural networks (ANNs) used in machine learning.
Key directions include:
- Dynamics of biological networks: investigating the dynamics of biochemical networks involved in signal transduction, adaptation and synchronization, as well as neural networks responsible for encoding, computation and learning.
- Inference of biological networks: reverse engineering the underlying biochemical networks (structure and dynamics) of functional biological networks from experimental data based on machine learning methods.
- Energetic costs of information processing: exploring how physical constraints on energy dissipation shape molecular mechanisms in biology and guide the design of efficient learning algorithms.
- Statistical physics of deep learning: aimed at understanding the mechanisms behind how ANNs learn and generalize and comparing these processes to learning and computation in real neural systems.
This research is carried out in collaboration with the Center for Computational Neuroscience, whose research focus on information processing in biological networks emphasizes biological and artificial neural networks.
