Prof. Dr. Sebastian Pokutta
Zuse Institute Berlin (ZIB)
Optimization and Machine Learning
Institute of Mathematics
Electrical Engineering and Computer Science (courtesy)
Technische Universität Berlin
Research Lab. My lab is interested in Artificial Intelligence, Optimization, and Machine Learning. We develop new methodologies (e.g., new optimization and learning algorithms), work on combining learning and decision-making, as well as design AI Systems for real-world deployment in various application contexts.
- Locally Accelerated Conditional Gradients
A. Carderera, J. Diakonikolas, S. Pokutta
preprint (2019), [arXiv] [summary] [slides]
- Blended Matching Pursuit
C. Combettes, S. Pokutta
to appear in Proceedings of NeurIPS (2019), [arXiv] [summary] [slides]
- Restarting Frank-Wolfe
T. Kerdreux, A. d’Aspremont, S. Pokutta
to appear in Proceedings of AISTATS (2019), [arXiv] [summary] [slides]
- Principled Deep Neural Network Training through Linear Programming
D. Bienstock, G. Muñoz, S. Pokutta
preprint (2018), [arXiv] [summary]
- Blended Conditional Gradients: the unconditioning of conditional gradients
G. Braun, S. Pokutta, D. Tu, S. Wright
to appear in Proceedings of ICML (2019), [arXiv] [summary] [poster] [slides]
Select Recent Talks and Teaching.
- 09/2019: (technical) “Smooth Constraint Convex Minimization via Conditional Gradients”. Plenary at the 19th French-German-Swiss conference on Optimization. [slides]
- 09/2019: (technical) “Mirror Descent and related methods in Linear and Discrete Optimization” at Cargese Workshop on Combinatorial Optimization. [slides]
- 07/2019: (technical) “Locally Accelerated Conditional Gradients” at RIMS, Kyoto. [slides]
- 01/2019: (technical) “Smooth Constraint Convex Minimization via Conditional Gradients”. Plenary at INFORMS Computing Society Conference. [slides]
- 11/2018: (technical) “Blended Conditional Gradients” at Oberwolfach (unfortunately the animations do not work; see [arXiv] for exact algorithm). [slides]
- WS/2019: Discrete Optimization and Machine Learning (Seminar).
Recent Blog Posts.
- 09/2019: SCIP x Raspberry Pi: SCIP on Edge
- 08/2019: Universal Portfolios: how to (not) get rich
- 08/2019: Toolchain Tuesday No. 6
- 07/2019: Conditional Gradients and Acceleration
- 06/2019: Cheat Sheet: Acceleration from First Principles
- 09/2019: We are organizing the third workshop on “Discrete Optimization and Machine Learning” in May 2020 at Kyoto University in Kyoto.
- 09/2019: TU Berlin and the Berlin School of Mathematics, with the support of MATH+, are organizing the “Combinatorial Optimization at Work (Co@Work) 2020” summer school on September 14 - 26, 2020 at ZIB in Berlin. Application deadline is: June 14, 2020. Intended audience: master/PhD students, Post-docs.
- 09/2019: Sanjeena Dang, Antoine Deza, Swati Gupta, Paul McNicholas, Masashi Sugiyama, and I are organizing a focus program on Data Science and Optimization in November 2019 at the Fields Institute in Toronto.
- 12/2018: Antoine Deza, Takanori Maehara, and I are editing a special issue on “Machine Learning and Discrete Optimization” in Discrete Optimization. Deadline for submission is May, 30th 2019.