Header Image

Robust and Explainable Learning

robustness and interpretability of artificial intelligence; compression of deep neural networks; optimization methods for deep learning; learning for combinatorial optimization

14
6
24

What we are interested in

Machine learning has revolutionised the process of analyzing data and has led to new insights and applications. However, one of the key short comings of this field is its use of black box models which may lack explainability. A key aspect of our research is to develop interpretable machine learning algorithms. We do so by using techniques such as adversarial optimization, sparsity, etc. We also focus on the interplay between learning and optimization by developing new algorithms to train machine learning models and exploting machine learning models to improve the process of optimization. Finally, we also aim to identify the theoretical reasons behind the success of such models.

🧑‍🎓 Members

Sebastian Pokutta
Department Head
pokutta (at) zib.de
Max Zimmer
Research Area Lead
zimmer (at) zib.de
Sai Ganesh Nagarajan
Research Area Lead
nagarajan (at) zib.de
Shpresim Sadiku
sadiku (at) zib.de
Berkant Turan
turan (at) zib.de
Ingo Meise
meise (at) zib.de
Moritz Wagner
wagner (at) zib.de
Konrad Mundinger
mundinger (at) zib.de
Christophe Roux
roux (at) zib.de
Christoph Graczyk
graczyk (at) zib.de
Kartikeya Chitranshi
chitranshi (at) zib.de
Nico Pelleriti
pelleriti (at) zib.de
Mihail Birsan
birsan (at) zib.de
Suhrab Asadulla
asadull (at) zib.de

🔬 Projects

AI-Based High-Resolution Forest Monitoring

Preserving global vegetation is crucial for addressing and mitigating climate change. Accurate, up-to-date forest health data is essential. AI4Forest aims to develop advanced AI methods to monitor forests using satellite imagery, including radar and optical data. The project will create scalable techniques for detailed, high-resolution maps of the globe, e.g., to monitor canopy height, biomass, and to track forest disturbances.

AI4Forest
Jun 2023 to May 2027
3
3
Decomposition Methods for Mixed-integer Optimal Control

In this project, we study domain decomposition approaches for optimal control in gas transport networks. Our goal is to couple space-time-domain decomposition with machine learning and mixed-integer programming. We will develop NeTI (Network Tearing and Interconnection), a data-driven and physics-informed algorithm combining mixed-integer nonlinear programming, surrogate model learning, and graph decomposition strategies.

TRR-154 A05
Jan 2022 to Jun 2026
3
2
Expanding Merlin-Arthur Classifiers: Interpretable Neural Networks Through Interactive Proof Systems

Existing approaches for interpreting Neural Network classifiers that highlight features relevant for a decision are based solely on heuristics. We introduce a theory that allows us to bound the quality of the features without assumptions on the classifier model by relating classification to Interactive Proof Systems.

MATH+ EF1-24
Apr 2023 to Mar 2026
4
3
Learning to Schedule Heuristics in IP

Heuristics play a crucial role in exact solvers for Mixed Integer Programming (MIP). However, the question of how to manage multiple MIP heuristics in a solver has not received sufficient attention. This project addresses the strategic management of primal heuristics in MIP solvers, aiming to replace static, hard-coded rules with dynamic, self-improving procedures.

HLEARN
Nov 2021 to Oct 2023
2
2
Adaptive Algorithms Through Machine Learning: Exploiting Interactions in Integer Programming

The performance of modern mixed-integer program solvers is highly dependent on a number of interdependent individual components. Using tools from machine learning, we intend to develop an integrated framework that is able to capture interactions of individual decisions made in these components with the ultimate goal to improve performance.

MATH+ EF1-9
Jan 2021 to Dec 2022
5
3
Globally Optimal Neural Network Training

Training artificial neural networks is a key optimization task in deep learning. To improve generalization, robustness, and explainability, we aim to compute globally optimal solutions. We will use integer programming methods, exploiting mixed-integer nonlinear programming and enhancing solving techniques like spatial branch-and-cut. Additionally, we'll leverage symmetry to reduce computational burden and ensure symmetry in solutions, and incorporate true sparsity using a mixed-integer nonlinear programming framework.

GONNT
Mar 2021 to Feb 2022
2

đź’¬ Talks and posters

Conference and workshop talks

Jul 2024
Unified Taxonomy in AI Safety: Watermarks, Adversarial Defenses, and Transferable Attacks by Berkant Turan
Workshop on Theoretical Foundations of Foundation Models (TF2M) @ ICML 2024, Vienna
Jul 2023
Extending Merlin-Arthur Classifiers for Improved Interpretability by Berkant Turan
The 1st World Conference on EXplainable Artificial Intelligence

Poster presentations

May 2024
Sparse Model Soups a Recipe for Improved Pruning Via Model Averaging by Max Zimmer
12th ICLR Conference, Vienna
May 2023
How I Learned to Stop Worrying and Love Retraining by Max Zimmer
11th ICLR Conference, Kigali

đź“ť Publications and preprints

  1. Deza, A., Pokutta, S., and Pournin, L. (2024). The Complexity of Geometric Scaling. Operations Research Letters, 52. DOI: 10.1016/j.orl.2023.11.010 [arXiv]
    [BibTeX]
    @article{2022_DezaPokuttaPournin_Geometricscaling,
  year = {2024},
  journal = {Operations Research Letters},
  volume = {52},
  doi = {10.1016/j.orl.2023.11.010},
  archiveprefix = {arXiv},
  eprint = {2205.04063},
  primaryclass = {math.OC},
  author = {Deza, Antoine and Pokutta, Sebastian and Pournin, Lionel},
  title = {The Complexity of Geometric Scaling}
}
  2. Wäldchen, S., Sharma, K., Turan, B., Zimmer, M., and Pokutta, S. (2024). Interpretability Guarantees with Merlin-Arthur Classifiers. Proceedings of International Conference on Artificial Intelligence and Statistics. [arXiv]
    [BibTeX]
    @inproceedings{2022_WaeldchenEtAl_Interpretabilityguarantees,
  year = {2024},
  booktitle = {Proceedings of International Conference on Artificial Intelligence and Statistics},
  archiveprefix = {arXiv},
  eprint = {2206.00759},
  primaryclass = {cs.LG},
  author = {Wäldchen, Stephan and Sharma, Kartikey and Turan, Berkant and Zimmer, Max and Pokutta, Sebastian},
  title = {Interpretability Guarantees with Merlin-Arthur Classifiers}
}
  3. Zimmer, M., Spiegel, C., and Pokutta, S. (2024). Sparse Model Soups: A Recipe for Improved Pruning Via Model Averaging. Proceedings of International Conference on Learning Representations. [URL] [arXiv]
    [BibTeX]
    @inproceedings{2023_ZimmerSpiegelPokutta_Sparsemodelsoups,
  year = {2024},
  booktitle = {Proceedings of International Conference on Learning Representations},
  url = {https://iclr.cc/virtual/2024/poster/17433},
  archiveprefix = {arXiv},
  eprint = {2306.16788},
  primaryclass = {cs.LG},
  author = {Zimmer, Max and Spiegel, Christoph and Pokutta, Sebastian},
  title = {Sparse Model Soups: A Recipe for Improved Pruning Via Model Averaging}
}
  4. Göß, A., Martin, A., Pokutta, S., and Sharma, K. (2024). Norm-induced Cuts: Optimization with Lipschitzian Black-box Functions. [URL] [arXiv]
    [BibTeX]
    @misc{2024_AdrianMartinPokuttaSharma_Norminducedcuts,
  url = {https://opus4.kobv.de/opus4-trr154/files/518/nic_preprint.pdf},
  archiveprefix = {arXiv},
  eprint = {2403.11546},
  primaryclass = {math.OC},
  year = {2024},
  author = {Göß, Adrian and Martin, Alexander and Pokutta, Sebastian and Sharma, Kartikey},
  title = {Norm-induced Cuts: Optimization with Lipschitzian Black-box Functions}
}
  5. Goerigk, M., Hartisch, M., Merten, S., and Sharma, K. (2024). Feature-Based Interpretable Optimization. [arXiv]
    [BibTeX]
    @misc{2024_GoerigkHartischMertenSharma_Featureinterpretableoptimization,
  archiveprefix = {arXiv},
  eprint = {2409.01869},
  primaryclass = {math.OC},
  year = {2024},
  author = {Goerigk, Marc and Hartisch, Michael and Merten, Sebastian and Sharma, Kartikey},
  title = {Feature-Based Interpretable Optimization}
}
  6. GĹ‚uch, G., Turan, B., Nagarajan, S. G., and Pokutta, S. (2024). The Good, the Bad and the Ugly: Watermarks, Transferable Attacks and Adversarial Defenses. [arXiv]
    [BibTeX]
    @misc{2024_GrzegorzTuranNagarajanPokutta_Watermarksadversarialdefenses,
  archiveprefix = {arXiv},
  eprint = {2410.08864},
  primaryclass = {cs.LG},
  year = {2024},
  author = {GĹ‚uch, Grzegorz and Turan, Berkant and Nagarajan, Sai Ganesh and Pokutta, Sebastian},
  title = {The Good, the Bad and the Ugly: Watermarks, Transferable Attacks and Adversarial Defenses}
}
  7. Mundinger, K., Zimmer, M., and Pokutta, S. (2024). Neural Parameter Regression for Explicit Representations of PDE Solution Operators. [arXiv]
    [BibTeX]
    @misc{2024_MundingerZimmerPokutta_Neuralparameterregression,
  archiveprefix = {arXiv},
  eprint = {2403.12764},
  primaryclass = {cs.LG},
  year = {2024},
  author = {Mundinger, Konrad and Zimmer, Max and Pokutta, Sebastian},
  title = {Neural Parameter Regression for Explicit Representations of PDE Solution Operators}
}
  8. Bienstock, D., Muñoz, G., and Pokutta, S. (2023). Principled Deep Neural Network Training Through Linear Programming. Discrete Optimization. [URL] [arXiv] [summary]
    [BibTeX]
    @article{2018_DanielGonzaloPokutta_Polyhedraltraining,
  year = {2023},
  journal = {Discrete Optimization},
  url = {https://www.sciencedirect.com/science/article/abs/pii/S1572528623000373},
  archiveprefix = {arXiv},
  eprint = {1810.03218},
  primaryclass = {cs.LG},
  author = {Bienstock, Daniel and Muñoz, Gonzalo and Pokutta, Sebastian},
  title = {Principled Deep Neural Network Training Through Linear Programming},
  summary = {https://www.pokutta.com/blog/research/2018/10/12/DNN-learning-lp-abstract.html}
}
  9. Zimmer, M., Spiegel, C., and Pokutta, S. (2023). How I Learned to Stop Worrying and Love Retraining. Proceedings of International Conference on Learning Representations. [URL] [arXiv] [code]
    [BibTeX]
    @inproceedings{2021_ZimmerSpiegelPokutta_Retrainingpruning,
  year = {2023},
  booktitle = {Proceedings of International Conference on Learning Representations},
  url = {https://iclr.cc/virtual/2023/poster/10914},
  archiveprefix = {arXiv},
  eprint = {2111.00843},
  primaryclass = {cs.LG},
  author = {Zimmer, Max and Spiegel, Christoph and Pokutta, Sebastian},
  title = {How I Learned to Stop Worrying and Love Retraining},
  code = {https://github.com/ZIB-IOL/BIMP}
}
  10. Kevin-Martin, A., Bärmann, A., Braun, K., Liers, F., Pokutta, S., Schneider, O., Sharma, K., and Tschuppik, S. (2023). Data-driven Distributionally Robust Optimization Over Time. INFORMS Journal on Optimization. [arXiv]
    [BibTeX]
    @article{2023_KevinEtAl_Datadrivendistributionallyrobust,
  year = {2023},
  journal = {INFORMS Journal on Optimization},
  archiveprefix = {arXiv},
  eprint = {2304.05377},
  primaryclass = {math.OC},
  author = {Kevin-Martin, Aigner and Bärmann, Andreas and Braun, Kristin and Liers, Frauke and Pokutta, Sebastian and Schneider, Oskar and Sharma, Kartikey and Tschuppik, Sebastian},
  title = {Data-driven Distributionally Robust Optimization Over Time}
}
  11. Kruser, J., Sharma, K., Holl, J., and Nohadani, O. (2023). Identifying Patterns of Medical Intervention in Acute Respiratory Failure: A Retrospective Observational Study. Critical Care Explorations.
    [BibTeX]
    @article{2023_KruserSharmaHollNohadani_Medicalinterventionpatterns,
  year = {2023},
  journal = {Critical Care Explorations},
  author = {Kruser, Jacqueline and Sharma, Kartikey and Holl, Jane and Nohadani, Omid},
  title = {Identifying Patterns of Medical Intervention in Acute Respiratory Failure: A Retrospective Observational Study}
}
  12. Sadiku, S., Wagner, M., and Pokutta, S. (2023). Group-wise Sparse and Explainable Adversarial Attacks. [arXiv]
    [BibTeX]
    @misc{2023_SadikuWagnerPokutta_Groupwisesparseattacks,
  archiveprefix = {arXiv},
  eprint = {2311.17434},
  primaryclass = {cs.CV},
  year = {2023},
  author = {Sadiku, Shpresim and Wagner, Moritz and Pokutta, Sebastian},
  title = {Group-wise Sparse and Explainable Adversarial Attacks}
}
  13. Thuerck, D., Sofranac, B., Pfetsch, M., and Pokutta, S. (2023). Learning Cuts Via Enumeration Oracles. Proceedings of Conference on Neural Information Processing Systems. [arXiv]
    [BibTeX]
    @inproceedings{2023_ThuerckSofranacPfetschPokutta_Learningcutsenumeration,
  year = {2023},
  booktitle = {Proceedings of Conference on Neural Information Processing Systems},
  archiveprefix = {arXiv},
  eprint = {2305.12197},
  primaryclass = {math.OC},
  author = {Thuerck, Daniel and Sofranac, Boro and Pfetsch, Marc and Pokutta, Sebastian},
  title = {Learning Cuts Via Enumeration Oracles}
}
  14. Zimmer, M., Andoni, M., Spiegel, C., and Pokutta, S. (2023). PERP: Rethinking the Prune-Retrain Paradigm in the Era of LLMs. [arXiv] [code]
    [BibTeX]
    @misc{2023_ZimmerAndoniSpiegelPokutta_PerpPruneRetrain,
  archiveprefix = {arXiv},
  eprint = {2312.15230},
  primaryclass = {cs.LG},
  year = {2023},
  author = {Zimmer, Max and Andoni, Megi and Spiegel, Christoph and Pokutta, Sebastian},
  title = {PERP: Rethinking the Prune-Retrain Paradigm in the Era of LLMs},
  code = {https://github.com/ZIB-IOL/PERP}
}
  15. Macdonald, J., Besançon, M., and Pokutta, S. (2022). Interpretable Neural Networks with Frank-Wolfe: Sparse Relevance Maps and Relevance Orderings. Proceedings of International Conference on Machine Learning. [arXiv] [poster] [video]
    [BibTeX]
    @inproceedings{2021_MacdonaldBesanconPokutta_Interpretableneuralnetworks,
  year = {2022},
  booktitle = {Proceedings of International Conference on Machine Learning},
  archiveprefix = {arXiv},
  eprint = {2110.08105},
  primaryclass = {cs.lG},
  author = {Macdonald, Jan and Besançon, Mathieu and Pokutta, Sebastian},
  title = {Interpretable Neural Networks with Frank-Wolfe: Sparse Relevance Maps and Relevance Orderings},
  poster = {https://pokutta.com/slides/20220712_icml_poster_interpretable_rde.pdf},
  video = {https://slideslive.com/38983588}
}
  16. Tsuji, K., Tanaka, K., and Pokutta, S. (2022). Pairwise Conditional Gradients without Swap Steps and Sparser Kernel Herding. Proceedings of International Conference on Machine Learning. [arXiv] [summary] [slides] [code] [video]
    [BibTeX]
    @inproceedings{2021_TsujiTanakaPokutta_Pairwiseconditionalgradients,
  year = {2022},
  booktitle = {Proceedings of International Conference on Machine Learning},
  archiveprefix = {arXiv},
  eprint = {2110.12650},
  primaryclass = {math.OC},
  author = {Tsuji, Kazuma and Tanaka, Ken'ichiro and Pokutta, Sebastian},
  title = {Pairwise Conditional Gradients without Swap Steps and Sparser Kernel Herding},
  code = {https://github.com/ZIB-IOL/FrankWolfe.jl},
  slides = {https://pokutta.com/slides/20220624_ICML2022_BPCG.pdf},
  summary = {https://pokutta.com/blog/research/2022/05/21/bpcg-abstract.html},
  video = {https://slideslive.com/38983561}
}
  17. Kossen, T., Hirzel, M. A., Madai, V. I., Boenisch, F., Hennemuth, A., Hildebrand, K., Pokutta, S., Sharma, K., Hilbert, A., Sobesky, J., Galinovic, I., Khalil, A. A., Fiebach, J. B., and Frey, D. (2022). Towards Sharing Brain Images: Differentially Private TOF-MRA Images with Segmentation Labels Using Generative Adversarial Networks. Frontiers in Artificial Intelligence. DOI: 10.3389/frai.2022.813842
    [BibTeX]
    @article{2022_KossenEtAl_Brainimagesharing,
  year = {2022},
  journal = {Frontiers in Artificial Intelligence},
  doi = {10.3389/frai.2022.813842},
  author = {Kossen, Tabea and Hirzel, Manuel A. and Madai, Vince I. and Boenisch, Franziska and Hennemuth, Anja and Hildebrand, Kristian and Pokutta, Sebastian and Sharma, Kartikey and Hilbert, Adam and Sobesky, Jan and Galinovic, Ivana and Khalil, Ahmed A. and Fiebach, Jochen B. and Frey, Dietmar},
  title = {Towards Sharing Brain Images: Differentially Private TOF-MRA Images with Segmentation Labels Using Generative Adversarial Networks}
}
  18. Nohadani, O., and Sharma, K. (2022). Optimization Under Connected Uncertainty. INFORMS Journal on Optimization. DOI: 10.1287/ijoo.2021.0067
    [BibTeX]
    @article{2022_NohadaniSharma_Connecteduncertainty,
  year = {2022},
  journal = {INFORMS Journal on Optimization},
  doi = {10.1287/ijoo.2021.0067},
  author = {Nohadani, Omid and Sharma, Kartikey},
  title = {Optimization Under Connected Uncertainty}
}
  19. Wäldchen, S., Huber, F., and Pokutta, S. (2022). Training Characteristic Functions with Reinforcement Learning: XAI-methods Play Connect Four. Proceedings of International Conference on Machine Learning. [arXiv] [poster] [video]
    [BibTeX]
    @inproceedings{2022_WaeldchenHuberPokutta_CharacteristicfunctionsReinforcementlearning,
  year = {2022},
  booktitle = {Proceedings of International Conference on Machine Learning},
  archiveprefix = {arXiv},
  eprint = {2202.11797},
  primaryclass = {cs.LG},
  author = {Wäldchen, Stephan and Huber, Felix and Pokutta, Sebastian},
  title = {Training Characteristic Functions with Reinforcement Learning: XAI-methods Play Connect Four},
  poster = {https://pokutta.com/slides/20220712_icml_poster_conn4.pdf},
  video = {https://slideslive.com/38983111}
}
  20. Zimmer, M., Spiegel, C., and Pokutta, S. (2022). Compression-aware Training of Neural Networks Using Frank-Wolfe. [arXiv]
    [BibTeX]
    @misc{2022_ZimmerSpiegelPokutta_Compressionawaretraining,
  archiveprefix = {arXiv},
  eprint = {2205.11921},
  primaryclass = {cs.LG},
  year = {2022},
  author = {Zimmer, Max and Spiegel, Christoph and Pokutta, Sebastian},
  title = {Compression-aware Training of Neural Networks Using Frank-Wolfe}
}
  21. Ziemke, T., Sering, L., Vargas Koch, L., Zimmer, M., Nagel, K., and Skutella, M. (2021). Flows Over Time As Continuous Limits of Packet-based Network Simulations. Transportation Research Procedia, 52, 123–130. DOI: 10.1016/j.trpro.2021.01.014 [URL]
    [BibTeX]
    @article{2021_ZiemkeEtAl_Flowscontinuouslimits,
  year = {2021},
  journal = {Transportation Research Procedia},
  volume = {52},
  pages = {123-130},
  doi = {10.1016/j.trpro.2021.01.014},
  url = {https://sciencedirect.com/science/article/pii/S2352146521000284},
  author = {Ziemke, Theresa and Sering, Leon and Vargas Koch, Laura and Zimmer, Max and Nagel, Kai and Skutella, Martin},
  title = {Flows Over Time As Continuous Limits of Packet-based Network Simulations}
}
  22. Combettes, C., Spiegel, C., and Pokutta, S. (2020). Projection-free Adaptive Gradients for Large-scale Optimization. [arXiv] [summary] [code]
    [BibTeX]
    @misc{2020_CombettesSpiegelPokutta_Projectionfreeadaptivegradients,
  archiveprefix = {arXiv},
  eprint = {2009.14114},
  primaryclass = {math.OC},
  year = {2020},
  author = {Combettes, Cyrille and Spiegel, Christoph and Pokutta, Sebastian},
  title = {Projection-free Adaptive Gradients for Large-scale Optimization},
  code = {https://github.com/ZIB-IOL/StochasticFrankWolfe},
  summary = {https://pokutta.com/blog/research/2020/10/21/adasfw.html}
}
  23. Pokutta, S., Spiegel, C., and Zimmer, M. (2020). Deep Neural Network Training with Frank-Wolfe. [arXiv] [summary] [code]
    [BibTeX]
    @misc{2020_PokuttaSpiegelZimmer_Frankwolfeneuralnetworks,
  archiveprefix = {arXiv},
  eprint = {2010.07243},
  primaryclass = {cs.LG},
  year = {2020},
  author = {Pokutta, Sebastian and Spiegel, Christoph and Zimmer, Max},
  title = {Deep Neural Network Training with Frank-Wolfe},
  code = {https://github.com/ZIB-IOL/StochasticFrankWolfe},
  summary = {https://pokutta.com/blog/research/2020/11/11/NNFW.html}
}
  24. Ziemke, T., Sering, L., Vargas Koch, L., Zimmer, M., Nagel, K., and Skutella, M. (2020). Flows Over Time As Continuous Limits of Packet-based Network Simulations. Proceedings of EURO Working Group on Transportation Meeting.
    [BibTeX]
    @inproceedings{2021_ZiemkeEtAl_Flowscontinuouslimits:1,
  year = {2020},
  booktitle = {Proceedings of EURO Working Group on Transportation Meeting},
  author = {Ziemke, Theresa and Sering, Leon and Vargas Koch, Laura and Zimmer, Max and Nagel, Kai and Skutella, Martin},
  title = {Flows Over Time As Continuous Limits of Packet-based Network Simulations}
}