Print bio

Professor Marcus Hutter

Marcus Hutter, Ph.D., Habil, is Senior Researcher at DeepMind and Honorary Professor in the Research School of Computer Science (RSCS) at the Australian National University (ANU) in Canberra.

At DeepMind since 2019, Marcus is researching and working on the Mathematical Foundations of Artificial General Intelligence.

Since 2000, his research interests are centered around the information-theoretic foundations of inductive reasoning, Reinforcement Learning, Artificial intelligence, Bayesian statistics, theoretical computer science, machine learning, sequential decision theory, universal forecasting, algorithmic information theory, adaptive control, MDL, image processing, particle physics, and philosophy of science. All this resulted in 300+ publications and several awards.

Marcus worked as an active software developer for various companies in several areas for many years, before he commenced his academic career in 2000 at the Artificial Intelligence (AI) institute IDSIA in Lugano, Switzerland, where he stayed for six years.

Between 1996 and 2000, Marcus was Software Developer and Project Leader at BrainLAB where he worked on Numerical Algorithms in the medical field.

In 2000, he joined Jürgen Schmidhuber’s group at IDSIA, the Istituto Dalle Molle di Studi sull’Intelligenza Artificiale in Manno, Switzerland. In 2002, Marcus, with Jürgen Schmidhuber and Shane Legg, developed and published a mathematical theory of artificial general intelligence, AIXI, based on idealized intelligent agents and reward-motivated reinforcement learning. His research at IDSIA was centered around the AIXI model, which is a mathematical top-down approach to AI, related to Kolmogorov complexity, algorithmic probability, universal Solomonoff induction, Occam’s razor, Levin search, sequential decision theory, dynamic programming, reinforcement learning, and rational agents.

In 2003 and 2004, he was also Honorary Lecturer at Technical University Munich on Algorithmic information theory and machine learning.

After leaving IDSIA in 2006, he became Associate Professor in the Research School of Information Sciences and Engineering (RSISE) at the Australian National University (ANU) in Canberra and Senior Researcher at the National Information and Communication Technology of Australia (NICTA).

In 2011, he took a Sabbatical Year in the Machine Learning Laboratory at the ETHZ and has been Full Professor at RSCS since 2011. For two years, between 2014 and 2016, Marcus was also Associate Director of Research at [email protected]

His book Universal Artificial Intelligence: Sequential Decisions Based on Algorithmic Probability was published by Springer in 2005. In this book, he develops a parameter-free theory of an optimal reinforcement learning agent embedded in an arbitrary unknown environment, based on a formal mathematical definition of general intelligence. He also authored Fitness Uniform Selection to Preserve Genetic Diversity, Instantons in QCD: Theory and Application of the Instanton Liquid Model, Robust Estimators under the Imprecise Dirichlet Model, and The Fastest and Shortest Algorithm for All Well-Defined Problems.

He coauthored Hybrid Rounding Techniques for Knapsack Problems, Adaptive Online Prediction by Following the Perturbed Leader, Asymptotics of Discrete MDL for Online Prediction, Distribution of Mutual Information from Complete and Incomplete Data, Optimality of Universal Bayesian Sequence Prediction for General Loss and Alphabet, and Family Structure from Periodic Solutions of an Improved Gap Equation. Read his full list of publications! Learn about his lectures.

Marcus runs the 50,000 € Hutter Prize for Compressing Human Knowledge. Launched in 2006, with 50,000 euros total funding and since 2020 with 500,000 Euros. The prize awards 5000 euros for each one percent improvement in the compressed size of the file enwik9, which is the larger of two files used in the Large Text Compression Benchmark. enwik9 consists of the first 1,000,000,000 characters of a specific version of English Wikipedia. The ongoing competition is organized by Marcus, Matt Mahoney, and Jim Bowery. In 2020, Marcus launched the 500,000 € Hutter Prize.

The KurzweilAI.net article Hutter Prize for Lossless Compression of Human Knowledge said

Marcus Hutter has announced the 50,000 Euro Hutter Prize for Lossless Compression of Human Knowledge by compressing the 100MB file Wikipedia ‘enwik8’ file to less than the current record of 18MB.

The intent of this prize is to encourage the development of intelligent compressors/programs.

“Being able to compress well is closely related to intelligence,” says the Prize for Compressing Human Knowledge” website.

“While intelligence is a slippery concept, file sizes are hard numbers. Wikipedia is an extensive snapshot of Human Knowledge. If you can compress the first 100MB of Wikipedia better than your predecessors, you(r compressor) likely has to be smart(er).”

The goal of the Hutter Prize is to encourage research in artificial intelligence (AI). The organizers believe that text compression and AI are equivalent problems. Marcus proved that the optimal behavior of a goal-seeking agent in an unknown but computable environment is to guess at each step that the environment is probably controlled by one of the shortest programs consistent with all interactions so far.

Lifeboat Foundation

Safeguarding Humanity

Advisory Board

Professor Marcus Hutter

Member of