Wojciech Czarnecki
Senior Research Scientist at DeepMind
About me

Scientist, gamer and an AI enthusiast living in London, UK. Currently focusing on multi-agent deep reinforcement learning as a path to develop general artificial intelligence.

Recent project
AlphaStar - Mastering the game of Starcraft II
The first AI to ever win against professional player in a full game of SC II
Resume
Experience
2017 - Present
Senior Research Scientist
DeepMind

Solving Intelligence

2016 - 2017
Research Engineer
DeepMind

Solving Intelligence

2014 - 2017
Research and Teaching Assistant
Jagiellonian University

2014 - 2014
Machine Learning Engineer
Samsung R&D

Education
2011 - 2015
Doctor of Philosophy
Machine Learning
Jagiellonian University

"Renyi's quadratic entropy in classification" (with distinction)

2006 - 2011
Master of Science
Theoretical Computer Science
Jagiellonian University

"Automated reconstruction of three-dimensional neuron structure from confocal microscopy image stack based on artificial neural networks"

Papers
2019
Science
Human-level performance in 3D multiplayer games with population-based reinforcement learning
Jaderberg Max, Czarnecki Wojciech M, Dunning Iain, Marris Luke, Lever Guy, Castaneda Antonio Garcia, Beattie Charles, Rabinowitz Neil C, Morcos Ari S, Ruderman Avraham, others
ICML 2019
Open-ended Learning in Symmetric Zero-sum Games
Balduzzi David, Garnelo Marta, Bachrach Yoram, Czarnecki Wojciech M, Perolat Julien, Jaderberg Max, Graepel Thore
AISTATS 2019
Distilling Policy Distillation
Czarnecki Wojciech Marian, Pascanu Razvan, Osindero Simon, Jayakumar Siddhant M, Swirszcz Grzegorz, Jaderberg Max
arXiv preprint arXiv:1903.01373
α-Rank: Multi-Agent Evaluation by Evolution
Omidshafiei Shayegan, Papadimitriou Christos, Piliouras Georgios, Tuyls Karl, Rowland Mark, Lespiau Jean-Baptiste, Czarnecki Wojciech M, Lanctot Marc, Perolat Julien, Munos Remi
ICLR 2019 Workshop on “Structure & Priors in Reinforcement Learning” (SPiRL)
Perception-Prediction-Reaction Agents for Deep Reinforcement Learning
Stooke Adam, Dalibard Valentin, Jayakumar Siddhant M, Czarnecki Wojciech M, Jaderberg Max
Blog post
AlphaStar: Mastering the Real-Time Strategy Game StarCraft II
Vinyals O, Babuschkin I, Chung J, Mathieu M, Jaderberg M, Czarnecki W, Dudzik A, Huang A, Georgiev P, Powell R, others
2018
NIPS 2018 DL Workshop
Kickstarting Deep Reinforcement Learning
Schmitt Simon, Hudson Jonathan J, Zidek Augustin, Osindero Simon, Doersch Carl, Czarnecki Wojciech M, Leibo Joel Z, Kuttler Heinrich, Zisserman Andrew, Simonyan Karen, others
ICML 2018
Progress & Compress: A scalable framework for continual learning
Schwarz Jonathan, Luketina Jelena, Czarnecki Wojciech M, Grabska-Barwinska Agnieszka, Teh Yee Whye, Pascanu Razvan, Hadsell Raia
Molecules
Substructural Connectivity Fingerprint and Extreme Entropy Machines—A New Method of Compound Representation and Analysis
Rataj Krzysztof, Czarnecki Wojciech , Podlewska Sabina, Pocha Agnieszka, Bojarski Andrzej
ICML 2018
Mix&Match-Agent Curricula for Reinforcement Learning
Czarnecki Wojciech Marian, Jayakumar Siddhant M, Jaderberg Max, Hasenclever Leonard, Teh Yee Whye, Osindero Simon, Heess Nicolas, Pascanu Razvan
AAAI 2019
Multi-task deep reinforcement learning with popart
Hessel Matteo, Soyer Hubert, Espeholt Lasse, Czarnecki Wojciech , Schmitt Simon, van Hasselt Hado
AAMAS 2019
Evolving intrinsic motivations for altruistic behavior
Wang Jane X, Hughes Edward, Fernando Chrisantha, Czarnecki Wojciech M, Duenez-Guzman Edgar A, Leibo Joel Z
NeurIPS 2018 MetaLearning Workshop
Adapting auxiliary losses using gradient similarity
Du Yunshu, Czarnecki Wojciech M, Jayakumar Siddhant M, Pascanu Razvan, Lakshminarayanan Balaji
Negotiating Team Formation Using Deep Reinforcement Learning
Bachrach Yoram, Everett Richard, Hughes Edward, Lazaridou Angeliki, Leibo Joel, Lanctot Marc, Johanson Mike, Czarnecki Wojtek, Graepel Thore
ICLR 2019
Information asymmetry in KL-regularized RL
Galashov Alexandre, Jayakumar Siddhant, Hasenclever Leonard, Tirumala Dhruva, Schwarz Jonathan, Desjardins Guillaume, Czarnecki Wojtek M, Teh Yee Whye, Pascanu Razvan, Heess Nicolas
2017
Pattern Analysis and Applications
Extreme entropy machines: robust information theoretic classification
Czarnecki Wojciech Marian, Tabor Jacek
ICML 2017
Decoupled neural interfaces using synthetic gradients
Jaderberg Max, Czarnecki Wojciech Marian, Osindero Simon, Vinyals Oriol, Graves Alex, Kavukcuoglu Koray
ICLR 2017
Reinforcement learning with unsupervised auxiliary tasks
Jaderberg Max, Mnih Volodymyr, Czarnecki Wojciech Marian, Schaul Tom, Leibo Joel Z, Silver David, Kavukcuoglu Koray
Bioorganic & medicinal chemistry letters
Quo vadis G protein-coupled receptor ligands? A tool for analysis of the emergence of new groups of compounds over time
Le\ sniak Damian, Jastrz\kebski Stanis\law, Podlewska Sabina, Czarnecki Wojciech M, Bojarski Andrzej J
Journal of chemical information and modeling
Creating the new from the old: Combinatorial libraries generation with machine-learning-based compound structure optimization
Podlewska Sabina, Czarnecki Wojciech M, Kafel Rafa\l, Bojarski Andrzej J
arXiv preprint arXiv:1702.02170
How to evaluate word embeddings? on importance of data efficiency and simple supervised tasks
Jastrzebski Stanis\law, Le\ sniak Damian, Czarnecki Wojciech Marian
TFML 2017
On loss functions for deep neural networks in classification
Janocha Katarzyna, Czarnecki Wojciech Marian
ICML 2017
Understanding Synthetic Gradients and Decoupled Neural Interfaces
Czarnecki Wojciech Marian, \ Swirszcz Grzegorz, Jaderberg Max, Osindero Simon, Vinyals Oriol, Kavukcuoglu Koray
NIPS 2017
Sobolev Training for Neural Networks
Czarnecki Wojciech Marian, Osindero Simon, Jaderberg Max, \ Swirszcz Grzegorz, Pascanu Razvan
AAMAS 2018
Value-decomposition networks for cooperative multi-agent learning
Sunehag Peter, Lever Guy, Gruslys Audrunas, Czarnecki Wojciech Marian, Zambaldi Vinicius, Jaderberg Max, Lanctot Marc, Sonnerat Nicolas, Leibo Joel Z, Tuyls Karl, others
NIPS 2017
Distral: Robust Multitask Reinforcement Learning
Teh Yee Whye, Bapst Victor, Czarnecki Wojciech Marian, Quan John, Kirkpatrick James, Hadsell Raia, Heess Nicolas, Pascanu Razvan
arXiv preprint arXiv:1711.09846
Population based training of neural networks
Jaderberg Max, Dalibard Valentin, Osindero Simon, Czarnecki Wojciech M, Donahue Jeff, Razavi Ali, Vinyals Oriol, Green Tim, Dunning Iain, Simonyan Karen, others
CoRR abs/1706.06551
Grounded language learning in a simulated 3d world
Hermann Karl Moritz, Hill Felix, Green Simon, Wang Fumin, Faulkner Ryan, Soyer Hubert, Szepesvari David, Czarnecki Wojciech Marian, Jaderberg Max, Teplyashin Denis, others
2016
ICLR 2016 Workshop track
Learning to SMILE (s)
Jastrzebski Stanislaw, Lesniak Damian, Czarnecki Wojciech Marian
Proceedings of the 9th International Conference on Computer Recognition Systems CORES 2015
Online extreme entropy machines for streams classification and active learning
Czarnecki Wojciech Marian, Tabor Jacek
arXiv preprint arXiv:1611.06310
Local minima in training of neural networks
Swirszcz Grzegorz, Czarnecki Wojciech Marian, Pascanu Razvan
GLISTEN Erlangen 2016 Conference
Teaching an old dog new tricks - optimized fingerprint as a basis for new compounds formation
Podlewska Sabina, Czarnecki Wojciech M, Kafel Rafa\l, Bojarski Andrzej J
2015
Bioorganic & medicinal chemistry letters
Exploiting uncertainty measures in compounds activity prediction using support vector machines
Smusz Sabina, Czarnecki Wojciech Marian, Warszycki Dawid, Bojarski Andrzej J
Schedae Informaticae
On the consistency of Multithreshold Entropy Linear Classifier
Czarnecki Wojciech Marian
Schedae Informaticae
Fast optimization of Multithreshold Entropy Linear Classifier
J\ ozefowicz Rafa\l, Czarnecki Wojciech Marian
Schedae Informaticae
Analysis of compounds activity concept learned by SVM using robust Jaccard based low-dimensional embedding
Jastrz\kebski Stanis\law, Czarnecki Wojciech Marian
Expert Systems with Applications
Multithreshold entropy linear classifier: Theory and applications
Czarnecki Wojciech Marian, Tabor Jacek
ECML PKDD 2015
Maximum Entropy Linear Manifold for Learning Discriminative Low-dimensional Representation
Czarnecki Wojciech Marian, J\ ozefowicz Rafa\l, Tabor Jacek
International Conference on Artificial Intelligence and Soft Computing
Adaptive active learning with ensemble of learners and multiclass problems
Czarnecki Wojciech Marian
IEEE Computational Intelligence Magazine
Weighted tanimoto extreme learning machine with case study in drug discovery
Czarnecki Wojciech Marian
Journal of Cheminformatics
Robust optimization of SVM hyperparameters in the classification of bioactive compounds
Czarnecki Wojciech M, Podlewska Sabina, Bojarski Andrzej
Trustcom/BigDataSE/ISPA 2015 IEEE
Compounds Activity Prediction in Large Imbalanced Datasets with Substructural Relations Fingerprint and EEM
Czarnecki Wojciech Marian, Rataj Krzysztof
ECML-PKDD 2015 2nd Workshop on Machine Learning in Life Sciences
Active Learning of Compounds Activity--Towards Scientifically Sound Simulation of Drug Candidates Identification
Czarnecki Wojciech Marian, Jastrzebski Stanisaw, Sieradzki Igor, Podlewska Sabina
Molecules
Extremely randomized machine learning methods for compound activity prediction
Czarnecki Wojciech , Podlewska Sabina, Bojarski Andrzej
Renyi s quadratic entropy in classification
Czarnecki Wojciech
2014
Man-Machine Interactions 3
InFeST--ImageJ Plugin for Rapid Development of Image Segmentation Pipelines
Czarnecki Wojciech Marian
Journal of Automation Mobile Robotics & Intelligent Systems
Designing a competition for autonomous robots with a restricted set of sensors with case study of LEGO NXT
Czarnecki Wojciech Marian, Szarzy\ nski Krzysztof, W\ ojtowicz Andrzej
Expert Systems with Applications
Two ellipsoid Support Vector Machines
Czarnecki Wojciech Marian, Tabor Jacek
arXiv preprint arXiv:1408.2869
Cluster based RBF Kernel for Support Vector Machines
Czarnecki Wojciech Marian, Tabor Jacek
GPCR Spring Conference 2014
Uncertainty of the in vitro experiments in the construction of predictive models
Smusz Sabina, Czarnecki Wojciech , Warszycki Dawid, Bojarski Andrzej J
Frontiers in Artificial Intelligence and Applications
Adaptive Active Learning as a Multi-armed Bandit Problem
Czarnecki Wojciech M., Podolak Igor T.
ECML-PKDD 2014 Neural Connectomics Workshop From Imaging to Connectivity
Neural Connectivity Reconstruction from Calcium Imaging Signal using Random Forest with Topological Features
Czarnecki WM, J\ ozefowicz R
Schedae Informaticae
Fast optimization of multithreshold entropy linear classifier
Rafa\l J\
2013
IFIP International Conference on Computer Information Systems and Industrial Management
Machine learning with known input data uncertainty measure
Czarnecki Wojciech M, Podolak Igor T
2012
Artificial Intelligence and Soft Computing
Multilayer Neural Networks with Receptive Fields as a Model for the Neuron Reconstruction Problem
Czarnecki Wojciech
Projects
AlphaStar - Mastering the game of Starcraft II
The first AI to ever win against professional player in a full game of SC II
Capture the Flag - the emergence of complex cooperative agents
Agents learning to perceive, strategise and cooperate with both other agents and humans in 3D CTF game
Get in Touch
  • Address . . . . . London, UK
  • Email . . . . . lejlot@
Meet me at the conference
November 2019
BlizzCon
Anaheim, USA

BlizzCon 2019 - Anaheim Convention Center

January 2020
Silesian Science Festival
Katowice, Poland

4th Silesian Science Festival

March 2020
SFI Academic IT Festival
Krakow, Poland

16th SFI Academic IT Festival