PeopleofSuperintelligence

Researcher

Manfred Opper

Manfred Opper is a machine-learning researcher whose work connects statistical physics with Bayesian inference and learning. In 1999 and 2000, he and Ole Winther develop Bayesian online learning, mean-field Gaussian-process classification, and a leave-one-out estimate of support vector machine generalization error. He co-edits Advanced Mean Field Methods: Theory and Practice in 2001 and develops expectation-consistent approximate inference in 2005. After a Reader post at Southampton, he accepts TU Berlin's call in artificial intelligence in 2006. As of October 2026, he is Professor Emeritus at TU Berlin and an ELLIS Fellow; his work includes stochastic control for Bayesian neural networks and inference for heterogeneous particle models.

Born
1956
Badges

Years link to their sources.

Timeline

  1. 1992

    A German physics award

    In 1992, he receives the Physics Award of the German Physical Society.

  2. 1999

    Bayesian online learning

    In 1999, he and Ole Winther present an online Bayesian algorithm that updates an approximate posterior as each example arrives. A Gaussian posterior family gives the algorithm asymptotic efficiency and supports learning in single-layer neural networks.

  3. 2000

    Gaussian processes and SVMs

    In 2000, he and Ole Winther present a mean-field method for Gaussian-process classification and a leave-one-out estimate of support vector machine generalization error.

  4. 2001

    An edited mean-field book

    He co-edits Advanced Mean Field Methods: Theory and Practice with David Saad in 2001; the book connects statistical-physics methods with approximate computation in probabilistic models.

  5. 2005

    Expectation-consistent inference

    With Ole Winther, he proposes expectation-consistent approximate inference for intractable probabilistic models in 2005, using a free-energy formulation and Gaussian approximations that retain correlations.

  6. 2005–2006

    A call to Berlin for AI

    In 2006, he accepts TU Berlin's call for its artificial-intelligence field after serving as a Reader at Southampton's School of Electronics and Computer Science.

  7. 2022

    Stochastic control for Bayesian neural networks

    In 2022, he coauthors Stochastic Control for Bayesian Neural Network Training with Ludwig Winkler and César Ojeda.

  8. As of October 2026

    Emeritus at TU Berlin

    As of October 2026, he is Professor Emeritus at TU Berlin and an ELLIS Fellow. In 2026, he coauthors a Communications Physics paper that estimates population heterogeneity in motile-particle trajectories with a likelihood approximation for second-order Langevin models.

Career

  1. Since 2006Technische Universität BerlinProfessor of Artificial Intelligence
  2. Until 2006University of SouthamptonReader

From the biography.

  1. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Technische Universität Berlin. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  2. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Technische Universität Berlin. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Technische Universität Berlin. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  4. Google has the highest combined h-index among AI organizationsIncludes Technische Universität Berlin. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Manfred Opper yet. Send us a free-licensed photo.

  1. Prof. Dr. Manfred Opper bifold.berlin. Checked 20 October 2026.
  2. A Bayesian approach to on-line learning — Aston Research Explorer research.aston.ac.uk. Checked 20 October 2026.
  3. Gaussian processes and SVM: Mean field results and leave-one-out — Aston Research Explorer research.aston.ac.uk. Checked 20 October 2026.
  4. Advanced mean field methods: theory and practice — Aston Research Explorer research.aston.ac.uk. Checked 20 October 2026.
  5. Expectation Consistent Approximate Inference jmlr.org. Checked 20 October 2026.
  6. Personalia der TU Berlin nachrichten.idw-online.de. Checked 20 October 2026.
  7. Honorary Colloquium for Manfred Opper sfb1294x.math.uni-potsdam.de. Checked 20 October 2026.
  8. Manfred Opper — ELLIS ellis.eu. Checked 20 October 2026.
  9. A Likelihood Approach for Inference of Population Heterogeneity in Particle Ensembles with Second-Order Langevin Dynamics nature.com. Checked 20 October 2026.