PeopleofSuperintelligence

Researcher

Patrik O. Hoyer

Also known as Patrik Hoyer, P.O. Hoyer, Patrick Hoyer

Computer scientist Patrik Hoyer is known for sparse non-negative data representations and for causal discovery methods such as LiNGAM. In 2002 he posted 'Non-negative sparse coding' on arXiv, and in 2004 the Journal of Machine Learning Research published his paper on non-negative matrix factorization with sparseness constraints. With Shohei Shimizu, Aapo Hyvärinen and Antti Kerminen he introduced the LiNGAM model in 2006, and in 2008 he coauthored a NeurIPS paper on nonlinear causal models with additive noise. At the University of Helsinki, he was a postdoc in 2007, an Academy Research Fellow at HIIT in 2010, and a coauthor of the 2009 Springer book Natural Image Statistics. In 2013 he coauthored 'Experiment Selection for Causal Discovery' and supervised Antti Hyttinen's computer science PhD.

Timeline

  1. 2002–2004

    Sparseness in matrix factorization

    In 2002, he posts 'Non-negative sparse coding' on arXiv, a method for decomposing multivariate data into non-negative sparse components. In 2004, the Journal of Machine Learning Research publishes his paper on non-negative matrix factorization with sparseness constraints, which adds explicit sparseness to the method and comes with MATLAB code.

  2. 2006

    The LiNGAM causal model

    He coauthors 'A Linear Non-Gaussian Acyclic Model for Causal Discovery' with Shohei Shimizu, Aapo Hyvärinen and Antti Kerminen in the Journal of Machine Learning Research in 2006. The LiNGAM method uses independent component analysis to discover the complete causal structure of continuous-valued data and comes with a MATLAB package.

  3. 2007

    Postdoc and a Carnegie Mellon visit

    In 2007, he is a postdoc at the Department of Computer Science of the University of Helsinki, working on machine learning methods related to independent component analysis and causal inference. He visits the Department of Philosophy at Carnegie Mellon University until December 2007.

  4. 2008

    Nonlinear additive noise models

    In 2008, he coauthors 'Nonlinear causal discovery with additive noise models' at NIPS with Dominik Janzing, Joris Mooij, Jonas Peters and Bernhard Schölkopf. The paper extends the linear framework to nonlinear models with additive noise, where nonlinearities provide information on the underlying causal system.

  5. 2008–2010

    Academy Research Fellow in Helsinki

    In December 2008, he is an academy research fellow at the University of Helsinki's Department of Computer Science and a supervisor in the Discovering Network Structures (DiNS) collaboration. In 2010, he is an Academy Research Fellow at HIIT and the University of Helsinki, among the supervisors of the Helsinki Graduate School in Computer Science and Engineering (Hecse).

  6. 2009

    A book and BayesLiNGAM

    In 2009, he coauthors Natural Image Statistics with Aapo Hyvärinen and Jarmo Hurri, a Springer-Verlag book on a probabilistic approach to early computational vision. In May 2009, he and Antti Hyttinen release the first public version of BayesLiNGAM, an R package for the Bayesian discovery of linear acyclic causal models.

  7. 2013

    Experiment selection for causal discovery

    In 2013, the Journal of Machine Learning Research publishes 'Experiment Selection for Causal Discovery', which he writes with Antti Hyttinen and Frederick Eberhardt and which gives explicit constructions of optimal sets of experiments. Hyttinen completes a PhD in computer science at the University of Helsinki in 2013 under Hoyer's supervision.

Career

  1. 2010Aalto UniversityAcademy Research Fellow
  2. 2007–2010University of HelsinkiPostdoc, Academy Research Fellow
  3. 2007Carnegie Mellon University

From the biography.

  1. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Carnegie Mellon University. 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.
  2. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Carnegie Mellon University. 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.
  3. Google has the highest combined h-index among AI organizationsIncludes Carnegie Mellon University. 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.
  4. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Carnegie Mellon University. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.

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 Patrik O. Hoyer yet. Send us a free-licensed photo.

  1. Non-negative sparse coding (arXiv:cs/0202009) arxiv.org. Checked 22 October 2026.
  2. Non-negative Matrix Factorization with Sparseness Constraints jmlr.org. Checked 22 October 2026.
  3. A Linear Non-Gaussian Acyclic Model for Causal Discovery jmlr.org. Checked 22 October 2026.
  4. Machine Learning and Friends at Carnegie Mellon: Causal discovery based on non-gaussianity cs.cmu.edu. Checked 22 October 2026.
  5. Nonlinear causal discovery with additive noise models papers.nips.cc. Checked 22 October 2026.
  6. PhD student positions in Helsinki on mining and learning networks k4all.org. Checked 22 October 2026.
  7. Hecse - Helsinki Graduate School in Computer Science and Engineering: Board of Hecse cs.helsinki.fi. Checked 22 October 2026.
  8. Natural Image Statistics cs.helsinki.fi. Checked 22 October 2026.
  9. Bayeslingam - Bayesian discovery of linear acyclic causal models cs.helsinki.fi. Checked 22 October 2026.
  10. Experiment Selection for Causal Discovery jmlr.org. Checked 22 October 2026.
  11. HIIT Alumni: Antti Hyttinen hiit.fi. Checked 22 October 2026.