Claudius Gros

Claudius Gros is a theoretical physicist whose research connects machine learning with neural computation and self-organized robots. He studied physics at ETH Zurich, pursued doctoral work in theoretical condensed-matter physics, and later held chairs in Saarland and Frankfurt. His research has included autonomously active neural networks, emotions as decision criteria, robot locomotion, and transformer task switching. As of October 2026, he is a professor at Goethe University Frankfurt working on complex systems theory, machine learning, and Physics of AI.

Now at: Goethe University Frankfurt
Works: 422
Badges: Loyal

Publication numbers from OpenAlex, as of 19 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

1961

Born in Mainz

He is born in Mainz in 1961.
1980s

Physics at ETH Zurich

After moving to Zurich in 1980, he studies physics at ETH Zurich and finishes in 1985. He stays at ETH for three years of PhD work in theoretical condensed-matter physics with T. M. Rice.
1988–2005

Academic posts in Germany

From 1988 to 1990, he conducts postdoctoral research at Indiana University with Steve Girvin and Allan MacDonald, then moves to Dortmund in 1990. He accepts a theoretical-physics chair at Saarland University in 1999 and moves to the University of Frankfurt for a chair in 2005.
2009

Autonomously active neural networks

His 2009 review examines how self-sustained neural activity can support cognitive computation, including attractor-relic networks that learn correlations between internal dynamics and sensory input.
2021

Emotion as a decision guide

In 2021, he argues that emotions act as abstract criteria for comparing activities and selecting goals and tasks in biological and artificial intelligences.
2024

Attractoring in robot locomotion

In 2024, he and Bulcsú Sándor review “attractoring,” a self-organized control principle in which feedback among controller, body, and environment produces locomotion in simulated and real-world robots.
2025

Transformers and cognitive science

In 2025, he examines small transformer architectures on task switching and compares them with recurrent networks and multilayer perceptrons. A separate 2025 article explores how world models, thought generation, attention, neural scaling, and quantization in machine learning may inform cognitive neuroscience.
As of October 2026

Physics of AI at Frankfurt

As of October 2026, he is a professor at Goethe University Frankfurt working in complex systems theory, machine learning, and Physics of AI, and teaches a Machine Learning Primer in winter 2026/27. His Physics of AI research analyzes attention as an information-routing mechanism in transformer models.