Henryk Michalewski

Google Research	henrykm@google.com
	henrykmichalewski@gmail.com
Oxford, UK	+44 750 825 6803

Education

2015: habilitation in computer science, University of Warsaw, title: Ivestigations of automata and related logics using methods of set theory
Winter 2002: internship in the Fields Institute, Toronto, Canada
1998-2002: PhD in Mathematics, University of Warsaw
1993-1998: MA in Mathematics, Univeristy of Warsaw, University of Warsaw

Work engagments

2021-: Staff Research Scientist at Google Brain
2021-: Associate Professor at the Department of Mathematics, Informatics and Mechanics of the University of Warsaw (on a leave)
2021-2022: Leverhulme Professor at the Department of Computer Science of the University of Oxford
2019-2021: Visiting Researcher at Google (Staff Faculty Visiting Researcher)
2018-2019: Visiting Professor at the Department of Computer Science of the University of Oxford
2017-2018: Assistant Professor at the Mathematical Institute of Polish Academy of Sciences (a temporary one year position)
2017: Invited Professor at the École normale supérieure de Lyon (one month position in the Laboratoire de l'Informatique du Parallélisme)
2016-2019: Data Scientist at deepsense.ai, responsible for research projects related to machine learning
2012-2013: Assistant Professor at the Mathematical Institute of Polish Academy of Sciences (a temporary one year position)
2007-2021: Assistant Professor at the Department of Mathematics, Informatics and Mechanics of the University of Warsaw
2004-2007: Postdoc at the Department of Mathematics of the Ben Gurion University, Israel. Working on topics related to logic and foundations of mathematics
2002-2004: Teaching Assistant at the Department of Mathematics, Informatics and Mechanics of the University of Warsaw

Selected papers 2018-22

Aitor Lewkowycz, Anders Andreassen, David Dohan, Ethan Dyer, Henryk Michalewski, Vinay Ramasesh, Ambrose Slone, Cem Anil, Imanol Schlag, Theo Gutman-Solo, Yuhuai Wu, Behnam Neyshabur, Guy Gur-Ari, Vedant Misra, Solving Quantitative Reasoning Problems with Language Models, blog post, sample explorer.
Kuang-Huei Lee, Ofir Nachum, Sherry Yang, Lisa Lee, Daniel Freeman, Winnie Xu, Sergio Guadarrama, Ian Fischer, Eric Jang, Henryk Michalewski, Igor Mordatch Multi-Game Decision Transformers, preprint 2022.
PaLM: Scaling Language Modeling with Pathways, preprint 2022. I contributed to Section 6.4 on Code Generation with Large Language Models.
Spyridon_Mouselinos, Henryk Michalewski, Mateusz Malinowski, Measuring CLEVRness: Black-box Testing of Visual Reasoning Models, ICLR 2022, .
Piotr Nawrot, Szymon Tworkowski, Michał Tyrolski, Łukasz Kaiser, Yuhuai Wu, Christian Szegedy, Henryk Michalewski, Hierarchical Transformers Are More Efficient Language Models, NAACL 2022.
Maxwell Nye, Anders Johan Andreassen, Guy Gur-Ari, Henryk Michalewski, Jacob Austin, David Bieber, David Dohan, Aitor Lewkowycz, Maarten Bosma, David Luan, Charles Sutton, Augustus Odena, Show Your Work: Scratchpads for Intermediate Computation with Language Models, preprint 2021.
Jacob Austin, Augustus Odena, Maxwell Nye, Maarten Bosma, Henryk Michalewski, David Dohan, Ellen Jiang, Carrie Cai, Michael Terry, Quoc Le, Charles Sutton, Program Synthesis with Large Language Models, preprint 2021.
Lukasz Kaiser, Mohammad Babaeizadeh, Piotr Milos, Blazej Osinski, Roy H Campbell, Konrad Czechowski, Dumitru Erhan, Chelsea Finn, Piotr Kozakowski, Sergey Levine, Afroz Mohiuddin, Ryan Sepassi, George Tucker, Henryk Michalewski, Model-Based Reinforcement Learning for Atari, spotlight at ICRL 2020 (5% acceptance rate).
Sebastian Jaszczur, Aakanksha Chowdhery, Afroz Mohiuddin, Lukasz Kaiser, Wojciech Gajewski, Jonni Kanerva, Henryk Michalewski, Green Transformers: Sparse is Enough, NeurIPS 2021.
Piotr Piekos, Henryk Michalewski, Mateusz Malinowski, Measuring and Improving BERT’s Mathematical Abilities by Predicting the Order of Reasoning, ACL 2021.
Cezary Kaliszyk, Josef Urban, Henryk Michalewski, Miroslav Olšák, Reinforcement learning of theorem proving, NeurIPS 2018.
Błażej Osiński, Adam Jakubowski, Paweł Zięcina, Piotr Miłoś, Christopher Galias, Silviu Homoceanu and Henryk Michalewski, Simulation-based reinforcement learning for real-world autonomous driving, ICRA 2020.
Sebastian Jaszczur, Michał Łuszczyk, Henryk Michalewski, Neural heuristics for SAT solving, poster and oral presentation at the Representation Learning on Graphs and Manifold Workshop at ICLR 2019.
Zsolt Zombori, Adrián Csiszárik, Henryk Michalewski, Cezary Kaliszyk, Josef Urban, Towards finding longer proofs, TABLEAUX 2021.

Other papers

Piotr Kozakowski, Lukasz Kaiser, Afroz Mohiuddin, Katarzyna Kanska, Henryk Michalewski, Q-Value Weighted Regression: Reinforcement Learning with Limited Data, preprint 2021.
Błażej Osiński, Piotr Miłoś, Adam Jakubowski, Paweł Zięcina, Michał Martyniak, Christopher Galias, Antonia Breuer, Silviu Homoceanu, Henryk Michalewski, CARLA Real Traffic Scenarios -- novel training ground and benchmark for autonomous driving, preprint 2021.
Igor Adamski, Robert Adamski, Tomasz Grel, Adam Jędrych, Kamil Kaczmarek, Henryk Michalewski, Distributed Deep Reinforcement Learning: Learn how to play Atari games in 21 minutes, presented at ISC HPC conference 2018.
Henryk Michalewski, Piotr Miłoś, Błażej Osiński, Learning to Run challenge solutions: Adapting reinforcement learning methods for neuromusculoskeletal environments, competition track at NIPS 2017, 6th place out of 450, some videos
Maciej Klimek, Henryk Michalewski, Piotr Miłoś, Hierarchical Reinforcement Learning with Parameters, presented at CoRL 2017, paper, videos
Robert Adamski, Tomasz Grel, Maciej Klimek, Henryk Michalewski: Atari games and Intel processors; CGW@IJCAI 2017, animations from various stages of learning for Breakout, Space Invaders and Boxing
Filippo Cavallari, Henryk Michalewski, Michał Skrzypczak: A characterisation of G_delta regular tree languages, MFCS 2017.
Mikolaj Bojańczyk, Henryk Michalewski: Some connections between universal algebra and logics for trees, preprint 2017
Henryk Michalewski,Matteo Mio, Michal Skrzypczak: Monadic Second Order Logic with Measure and Category Quantifiers, Logical Methods in Computer Science 14(2) (2018)
Jakub Sygnowski, Henryk Michalewski: Learning from the memory of Atari 2600, CGW@IJCAI 2016, implementation
Oskar Skibski, Henryk Michalewski, Andrzej Nagórko, Tomasz Michalak, Andrew Dowell, Talal Rahwan, Michael Wooldridge, Non-Utilitarian Coalition Structure Generation, ECAI 2016, implementation
Leszek Aleksander Kolodziejczyk, Henryk Michalewski, Pierre Pradic, Michal Skrzypczak: The Logical Strength of Büchi's Decidability Theorem, CSL 2016: 36:1-36:16
Leszek Aleksander Kolodziejczyk, Henryk Michalewski: How unprovable is Rabin's decidability theorem?, LICS 2016: 788-797
Henryk Michalewski, Andrzej Nagórko, Jakub Pawlewicz, An upper bound of 84 for Morpion Solitaire 5D, CCCG 2016 proceedings, implementation
Henryk Michalewski, Michal Skrzypczak: Unambiguous Büchi Is Weak, DLT 2016: 319-331
Henryk Michalewski, Matteo Mio, Mikolaj Bojańczyk: On the Regular Emptiness Problem of Subzero Automata, ICE 2016: 1-23
Henryk Michalewski, Matteo Mio, Measure Quantiﬁer in Monadic Second Order Logic, LFCS 2016: 267-282.
Henryk Michalewski, Matteo Mio, On the Problem of Computing the Probability of Regular Sets of Trees, FSTTCS 2015: 489-502
Henryk Michalewski, Matteo Mio. Baire Category Quantifier in Monadic Second Order Logic, ICALP (2) 2015: 362-374
Henryk Michalewski, Andrzej Nagórko, Jakub Pawlewicz, 485 - a new upper bound for Morpion Solitaire, CGW@IJCAI 2015, implementation
André Arnold, Henryk Michalewski, Damian Niwinski, On the Separation Question for Tree Languages, Theory Comput. Syst. 55(4): 833-855 (2014)
Alessandro Facchini, Henryk Michalewski, Deciding the Borel Complexity of Regular Tree Languages, CiE 2014: 163-172
Mikolaj Bojańczyk, Tomasz Gogacz, Henryk Michalewski, Michal Skrzypczak, On the Decidability of MSO+U on Infinite Trees, ICALP (2) 2014: 50-61
Tomasz Gogacz, Henryk Michalewski, Matteo Mio, Michal Skrzypczak, Measure Properties of Game Tree Languages, MFCS (1) 2014: 303-314
Henryk Michalewski, Damian Niwinski, On Topological Completeness of Regular Tree Languages, Logic and Program Semantics 2012: 165-179
André Arnold, Henryk Michalewski, Damian Niwinski, On the separation question for tree languages, STACS 2012: 396-407
Szczepan Hummel, Henryk Michalewski, Damian Niwinski, On the Borel Inseparability of Game Tree Languages, STACS 2009: 565-575
Menachem Kojman, Henryk Michalewski, Borel extensions of Baire measures in ZFC, Fund. Math. 211 (2011), no. 3, 197-223
Andrzej Komisarski, Henryk Michalewski, Pawel Milewski, Functions equivalent to Borel measurable ones, Bull. Pol. Acad. Sci. Math. 58 (2010), no. 1, 55-64
Adam Krawczyk, Witold Marciszewski, Henryk Michalewski, Remarks on the set of G_delta-points in Eberlein and Corson compact spaces, Topology Appl. 156 (2009), no. 9, 1746-1748.
Wieslaw Kubiś, Henryk Michalewski, Small Valdivia compact spaces, Topology and its Applications 153 (14), 2560-2573
Adam Krawczyk, Henryk Michalewski, An example of a topological group, Topology Appl. 127 (2003), no. 3, 325-330
Adam Krawczyk, Henryk Michalewski, Linear metric spaces close to being sigma–compact, preprint, 2003
Henryk Michalewski, Condensation of projective sets onto compact, Proc. AMS 131, (2003), 3601-3606
Henryk Michalewski, An answer to a question of Arhangelskii, Comment. Math. Univ. Carolin. 42 (2001), no. 3, 545-550
Henryk Michalewski, Homogeneity of K(Q), Tsukuba J. Math. 24 (2000), no. 2, 297-302
Henryk Michalewski, Game-theoretic approach to the hereditary Baire property of C_p(N_F), Bull. Polish Acad. Sci. Math. 46 (1998), no. 2, 135-140
Henryk Michalewski, Roman Pol, On a Hurewicz-type theorem and a selection theorem of Michael, Bull. Polish Acad. Sci. Math. 43 (1995), no. 4, 273-275 (1996)

Current Students

Spyridon Mouselinos, 2021-2025, Ph.D. project Towards visual reasoning, thesis co-supervised by Mateusz Malinowski (DeepMind).

Former Students

Pierre Pradic, 2016-2019, joint PhD program with the École normale supérieure de Lyon, thesis co-supervisied by Colin Riba; main topic: reverse mathematics of algorithms. Thesis defended at ENS in June 2020.
Błażej Osiński, 2017-2021, a former Googler, main topic: machine learning and robotics, currently working as a researcher at Lyft.
Adam Paszke, 2019-2020, BA thesis Automatic differentiation of invertible programs.

Recent grants and projects

Machine learning and automatic theorem proving, 2019-2022 Opus NCN grant, principal investigator.
Simulation-based reinforcement learning for real-world autonomous driving, 2018-2020, a joint project with Volkswagen.
Google Research Award dedicated to development of Google Research Football.
Several computation grants from plgrid and Google, along with some hardware deployed at the MIMUW computer lab obtained free of charge from Intel.

Prizes, invited talks

invited speaker at the Annual Meeting of the Swiss Society for Logic and Philosophy of Science (SSLPS), Lausanne, Switzerland, November 2015.
invited talk on reinforcement learning at the Google office in Warsaw, Poland, Feb 2018.
invited talk on reinforcement learning and theorem proving at Google Research, Mountain View, USA, November 2018.
invited talk on reinforcement learning and theorem proving at DeepMind, London, UK, July 2019.
invited talk on reinforcement learning and theorem proving at the Cambridge Computer Laboratory, UK, August 2019.
invited talk on reinforcement learning and theorem proving at the Cambridge Math Department, UK, November 2019 - an invitation from Tim Gowers.
invited talk on reinforcement learning and theorem proving at the Department of Computer Science of the University of Wroclaw, February 2020.

Program Committees

Teaching

machine learning: a course run jointly with Jan Lasek and Piotr Migdał from deepsense.ai for consultants of the Boston Consulting Group, including a kaggle-style machine learning comptetition organized on a machine learning platform neptune.ml
optimization: a novel course at the University of Warsaw on linear optimization combining precise exposition of theory and practical classes in programming
advanced courses at the University of Warsaw: practical machine learning run jointly with with deepsense.ai, computer aided verification, logic and type theory - programming in Coq, artificial intelligence and games (a research seminar, which led to 2 papers on optimization), automata on infinite structures, mu-calculus, information theory, algorithmic aspects of game theory, functional analysis, descriptive set theory, tropical geometry and linear optimization