Wouter Koolen-Wijkstra
- Full Name
- Dr. W.M. Koolen-Wijkstra
- Function(s)
- Scientific Staff Member
- W.M.Koolen-Wijkstra@cwi.nl
- Telephone
- +31 20 592 4009
- Room
- L132
- Department(s)
- Machine Learning
- Homepage
- http://wouterkoolen.info/
Biography
Wouter M. Koolen specialises in the foundations of sequential decision making. His contributions includes efficient adaptive learning algorithms that interpolate between adversarial and statistical models of learning, learning at the optimal rate in each case. His current research is focused on statistical and computational trade-offs in structured multi-armed bandit problems. Wouter is funded by a NWO Veni grant, and he is an inaugural member of the ACM Future of Computing Academy.Research
I work on topics in machine learning and related areas, including game theory, information theory, statistics and optimisation.
My current projects are
- Game tree search
- Learning faster from easy data
- Online learning beyond experts
- Minimax square loss prediction
- Online Finance
- Switching Experts
- Non-uniform regret analysis
- Robust statistics
- Algorithmic information theory
Publications
-
Kaufmann, E, Koolen-Wijkstra, W.M, & Garivier, A. (2018). Sequential test for the lowest mean: From Thompson to Murphy sampling. In Advances in Neural Information Processing Systems (pp. 6332–6342).
-
Kaufmann, E, & Koolen-Wijkstra, W.M. (2017). Monte-Carlo tree search by best arm identification. In Advances in Neural Information Processing Systems (pp. 4898–4907).
-
Kotłowski, W, Koolen-Wijkstra, W.M, & Malek, A. (2017). Random permutation online isotonic regression. In Advances in Neural Information Processing Systems 30 (NIPS 2017) (pp. 4181–4190).
-
Koolen-Wijkstra, W.M, Grünwald, P.D, & van Erven, T.A.L. (2016). Combining adversarial guarantees and stochastic fast rates in online learning. In Advances in Neural Information Processing Systems (pp. 4457–4465).
-
van Erven, T.A.L, & Koolen-Wijkstra, W.M. (2016). MetaGrad: multiple learning rates in online learning. In Advances in Neural Information Processing Systems (pp. 3666–3674).
-
Koolen-Wijkstra, W.M. (2016). Robust and adaptive methods for sequential decision making. ERCIM News, 107, 24–25.
-
Van Ommen, T. (Thijs), Koolen-Wijkstra, W.M, Feenstra, T.E, & Grünwald, P.D. (2016). Robust probability updating. International Journal of Approximate Reasoning, 74, 30–57. doi:10.1016/j.ijar.2016.03.001
-
Garivier, A, Kaufmann, E, & Koolen-Wijkstra, W.M. (2016). Maximin action identification: a new bandit framework for games. In V Feldman & A Rakhlin (Eds.), JMLR: Workshop and Conference Proceedings (pp. 1028–1050).
-
Kotlowski, W.T, Koolen-Wijkstra, W.M, & Malek, A. (2016). Online isotonic regression. In V Feldman & A Rakhlin (Eds.), JMLR: Workshop and Conference Proceedings (pp. 1165–1189).
-
Adamskyi, D, Koolen-Wijkstra, W.M, Chernov, A, & Vovk, V. (2016). A closer look at adaptive regret. Journal of Machine Learning Research, 17.
Professional activities
- Member: ACM Future of Computing Academy
Grants
- Veni Learning at the Intrinsic Task Pace 2015 (2015)
- Queensland University of Technology Vice-Chancellor's research fellowship The Versatile Multitask Learner (2013)
- Rubicon - Game-Theoretically Optimal Online Learning: From Conflicting Advice to High-Quality Decisions (2010)