Jovita Lukasik

Publications

2024

Conference paper

J. P. Schneider, M. Fatima, J. Lukasik, A. Kolb, M. Keuper, and M. Moeller

“Implicit Representations for Constrained Image Segmentation,” in Proceedings of the 41st International Conference on Machine Learning (ICML 2024), Vienna, Austria, 2024.

@inproceedings{convexSegICML2024,
TITLE = {Implicit Representations for Constrained Image Segmentation},
AUTHOR = {Schneider, Jan Philipp and Fatima, Mishal and Lukasik, Jovita and Kolb, Andreas and Keuper, Margret and Moeller, Michael},
LANGUAGE = {eng},
ISSN = {1938-7228},
URL = {https://proceedings.mlr.press/v235/},
PUBLISHER = {MLResearchPress},
YEAR = {2024},
DATE = {2024},
BOOKTITLE = {Proceedings of the 41st International Conference on Machine Learning (ICML 2024)},
EDITOR = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix},
PAGES = {43765--43790},
SERIES = {Proceedings of the Machine Learning Research},
VOLUME = {235},
ADDRESS = {Vienna, Austria},
}

Endnote

%0 Conference Proceedings
%A Schneider, Jan Philipp
%A Fatima, Mishal
%A Lukasik, Jovita
%A Kolb, Andreas
%A Keuper, Margret
%A Moeller, Michael
%+ External Organizations
External Organizations
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
External Organizations
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
External Organizations
%T Implicit Representations for Constrained Image Segmentation : 
%G eng
%U http://hdl.handle.net/21.11116/0000-000F-610A-9
%D 2024
%B 41at International Conference on Machine Learning
%Z date of event: 2024-07-21 - 2024-07-27
%C Vienna, Austria
%B Proceedings of the 41st International Conference on Machine Learning
%E Salakhutdinov, Ruslan; Kolter, Zico; Heller, Katherine; Weller, Adrian; Oliver, Nuria; Scarlett, Jonathan; Berkenkamp, Felix
%P 43765 - 43790
%I MLResearchPress
%U https://proceedings.mlr.press/v235/
%B Proceedings of the Machine Learning Research
%N 235
%@ false

2023

Conference paper

J. Lukasik, J. Geiping, M. Moeller, and M. Keuper

“Differentiable Architecture Search: a One-Shot Method?,” in AutoML Conference 2023, Potsdam/Berlin, Germany, 2023.

@inproceedings{lukasik2023differentiable,
TITLE = {Differentiable Architecture Search: a One-Shot Method?},
AUTHOR = {Lukasik, Jovita and Geiping, Jonas and Moeller, Michael and Keuper, Margret},
LANGUAGE = {eng},
URL = {https://openreview.net/forum?id=LV-5kHj-uV5; https://2023.automl.cc/},
PUBLISHER = {OpenReview.net},
YEAR = {2023},
MARGINALMARK = {$\bullet$},
BOOKTITLE = {AutoML Conference 2023},
PAGES = {1--18},
ADDRESS = {Potsdam/Berlin, Germany},
}

Endnote

%0 Conference Proceedings
%A Lukasik, Jovita
%A Geiping, Jonas
%A Moeller, Michael
%A Keuper, Margret
%+ External Organizations
External Organizations
External Organizations
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
%T Differentiable Architecture Search: a One-Shot Method?  : 
%G eng
%U http://hdl.handle.net/21.11116/0000-000F-2428-C
%U https://openreview.net/forum?id=LV-5kHj-uV5
%D 2023
%B International Conference on Automated Machine Learning
%Z date of event: 2023-09-12 - 2023-09-15
%C Potsdam/Berlin, Germany
%B AutoML Conference 2023
%P 1 - 18
%I OpenReview.net

Conference paper

S. Jung, J. Lukasik, and M. Keuper

“Neural Architecture Design and Robustness: A Dataset,” in Eleventh International Conference on Learning Representations (ICLR 2023), Kigali, Rwanda.

Abstract

Deep learning models have proven to be successful in a wide
range of machine learning tasks. Yet, they are often highly sensitive to
perturbations on the input data which can lead to incorrect decisions
with high confidence, hampering their deployment for practical
use-cases. Thus, finding architectures that are (more) robust against
perturbations has received much attention in recent years. Just like the
search for well-performing architectures in terms of clean accuracy,
this usually involves a tedious trial-and-error process with one
additional challenge: the evaluation of a network's robustness is
significantly more expensive than its evaluation for clean accuracy.
Thus, the aim of this paper is to facilitate better streamlined research
on architectural design choices with respect to their impact on
robustness as well as, for example, the evaluation of surrogate measures
for robustness. We therefore borrow one of the most commonly considered
search spaces for neural architecture search for image classification,
NAS-Bench-201, which contains a manageable size of 6466 non-isomorphic
network designs. We evaluate all these networks on a range of common
adversarial attacks and corruption types and introduce a database on
neural architecture design and robustness evaluations. We further
present three exemplary use cases of this dataset, in which we (i)
benchmark robustness measurements based on Jacobian and Hessian matrices
for their robustness predictability, (ii) perform neural architecture
search on robust accuracies, and (iii) provide an initial analysis of
how architectural design choices affect robustness. We find that
carefully crafting the topology of a network can have substantial impact
on its robustness, where networks with the same parameter count range in
mean adversarial robust accuracy from 20%-41%.

BibTeX

@inproceedings{Jung_ICLR23,
TITLE = {Neural Architecture Design and Robustness: {A} Dataset},
AUTHOR = {Jung, Steffen and Lukasik, Jovita and Keuper, Margret},
LANGUAGE = {eng},
PUBLISHER = {OpenReview.net},
YEAR = {2023},
PUBLREMARK = {Accepted},
MARGINALMARK = {$\bullet$},
ABSTRACT = {Deep learning models have proven to be successful in a wide <br>range of machine learning tasks. Yet, they are often highly sensitive to <br>perturbations on the input data which can lead to incorrect decisions <br>with high confidence, hampering their deployment for practical <br>use-cases. Thus, finding architectures that are (more) robust against <br>perturbations has received much attention in recent years. Just like the <br>search for well-performing architectures in terms of clean accuracy, <br>this usually involves a tedious trial-and-error process with one <br>additional challenge: the evaluation of a network's robustness is <br>significantly more expensive than its evaluation for clean accuracy. <br>Thus, the aim of this paper is to facilitate better streamlined research <br>on architectural design choices with respect to their impact on <br>robustness as well as, for example, the evaluation of surrogate measures <br>for robustness. We therefore borrow one of the most commonly considered <br>search spaces for neural architecture search for image classification, <br>NAS-Bench-201, which contains a manageable size of 6466 non-isomorphic <br>network designs. We evaluate all these networks on a range of common <br>adversarial attacks and corruption types and introduce a database on <br>neural architecture design and robustness evaluations. We further <br>present three exemplary use cases of this dataset, in which we (i) <br>benchmark robustness measurements based on Jacobian and Hessian matrices <br>for their robustness predictability, (ii) perform neural architecture <br>search on robust accuracies, and (iii) provide an initial analysis of <br>how architectural design choices affect robustness. We find that <br>carefully crafting the topology of a network can have substantial impact <br>on its robustness, where networks with the same parameter count range in <br>mean adversarial robust accuracy from 20%-41%.},
BOOKTITLE = {Eleventh International Conference on Learning Representations (ICLR 2023)},
ADDRESS = {Kigali, Rwanda},
}

Endnote

%0 Conference Proceedings
%A Jung, Steffen
%A Lukasik, Jovita
%A Keuper, Margret
%+ Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
%T Neural Architecture Design and Robustness: A Dataset : 
%G eng
%U http://hdl.handle.net/21.11116/0000-000C-738F-2
%D 2023
%B Eleventh International Conference on Learning Representations
%Z date of event: 2023-05-01 - 2023-05-05
%C Kigali, Rwanda
%X Deep learning models have proven to be successful in a wide <br>range of machine learning tasks. Yet, they are often highly sensitive to <br>perturbations on the input data which can lead to incorrect decisions <br>with high confidence, hampering their deployment for practical <br>use-cases. Thus, finding architectures that are (more) robust against <br>perturbations has received much attention in recent years. Just like the <br>search for well-performing architectures in terms of clean accuracy, <br>this usually involves a tedious trial-and-error process with one <br>additional challenge: the evaluation of a network's robustness is <br>significantly more expensive than its evaluation for clean accuracy. <br>Thus, the aim of this paper is to facilitate better streamlined research <br>on architectural design choices with respect to their impact on <br>robustness as well as, for example, the evaluation of surrogate measures <br>for robustness. We therefore borrow one of the most commonly considered <br>search spaces for neural architecture search for image classification, <br>NAS-Bench-201, which contains a manageable size of 6466 non-isomorphic <br>network designs. We evaluate all these networks on a range of common <br>adversarial attacks and corruption types and introduce a database on <br>neural architecture design and robustness evaluations. We further <br>present three exemplary use cases of this dataset, in which we (i) <br>benchmark robustness measurements based on Jacobian and Hessian matrices <br>for their robustness predictability, (ii) perform neural architecture <br>search on robust accuracies, and (iii) provide an initial analysis of <br>how architectural design choices affect robustness. We find that <br>carefully crafting the topology of a network can have substantial impact <br>on its robustness, where networks with the same parameter count range in <br>mean adversarial robust accuracy from 20%-41%.
%B Eleventh International Conference on Learning Representations
%I OpenReview.net

Conference paper

J. Lukasik, M. Moeller, and M. Keuper

“An Evaluation of Zero-Cost Proxies - From Neural Architecture Performance Prediction to Model Robustness,” in Pattern Recognition (DAGM GCPR 2023), Heidelberg, Germany, 2023.

@inproceedings{10.1007/978-3-031-54605-1_40,
TITLE = {An Evaluation of Zero-Cost Proxies -- From Neural Architecture Performance Prediction to Model Robustness},
AUTHOR = {Lukasik, Jovita and Moeller, Michael and Keuper, Margret},
LANGUAGE = {eng},
ISBN = {978-3-031-54604-4; 978-3-031-54605-1},
DOI = {10.1007/978-3-031-54605-1_40},
PUBLISHER = {Springer},
YEAR = {2023},
MARGINALMARK = {$\bullet$},
DATE = {2023},
BOOKTITLE = {Pattern Recognition (DAGM GCPR 2023)},
EDITOR = {K{\"o}the, Ullrich and Rother, Carsten},
PAGES = {624--638},
SERIES = {Lecture Notes in Computer Science},
VOLUME = {14264},
ADDRESS = {Heidelberg, Germany},
}

Endnote

%0 Conference Proceedings
%A Lukasik, Jovita
%A Moeller, Michael
%A Keuper, Margret
%+ External Organizations
External Organizations
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
%T An Evaluation of Zero-Cost Proxies - From Neural Architecture Performance Prediction to Model Robustness : 
%G eng
%U http://hdl.handle.net/21.11116/0000-000F-243C-6
%R 10.1007/978-3-031-54605-1_40
%D 2023
%B 45th German Conference on Pattern Recognition
%Z date of event: 2023-09-19 - 2023-09-22
%C Heidelberg, Germany
%B Pattern Recognition
%E K&#246;the, Ullrich; Rother, Carsten
%P 624 - 638
%I Springer
%@ 978-3-031-54604-4 978-3-031-54605-1
%B Lecture Notes in Computer Science
%N 14264
%U https://rdcu.be/dErn5

Article

J. Lukasik, P. Gavrikov, J. Keuper, and M. Keuper

“Improving Native CNN Robustness with Filter Frequency Regularization,” Transactions on Machine Learning Research, vol. 2023, 2023.

@article{lukasik2023improving,
TITLE = {Improving Native {CNN} Robustness with Filter Frequency Regularization},
AUTHOR = {Lukasik, Jovita and Gavrikov, Paul and Keuper, Janis and Keuper, Margret},
LANGUAGE = {eng},
ISSN = {2835-8856},
URL = {https://openreview.net/forum?id=2wecNCpZ7Y},
PUBLISHER = {TMLR},
ADDRESS = {New York, NY},
YEAR = {2023},
MARGINALMARK = {$\bullet$},
JOURNAL = {Transactions on Machine Learning Research},
VOLUME = {2023},
PAGES = {1--36},
}

Endnote

%0 Journal Article
%A Lukasik, Jovita
%A Gavrikov, Paul
%A Keuper, Janis
%A Keuper, Margret
%+ External Organizations
External Organizations
External Organizations
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
%T Improving Native CNN Robustness with Filter Frequency Regularization : 
%G eng
%U http://hdl.handle.net/21.11116/0000-000F-241E-8
%U https://openreview.net/forum?id=2wecNCpZ7Y
%7 2023
%D 2023
%J Transactions on Machine Learning Research
%V 2023
%& 1
%P 1 - 36
%I TMLR
%C New York, NY
%@ false
%U https://github.com/jovitalukasik/filter_freq_reg

Conference paper

J. P. Schneider, F. Mishal, J. Lukasik, A. Kolb, M. Keuper, and M. Moeller

“Implicit Representations for Image Segmentation,” in UniReps: The First Workshop on Unifying Representations in Neural Models, New Orleans, LA, USA, 2022.

@inproceedings{schneider2023implicit,
TITLE = {Implicit Representations for Image Segmentation},
AUTHOR = {Schneider, Jan Philipp and Mishal, Fatima and Lukasik, Jovita and Kolb, Andreas and Keuper, Margret and Moeller, Michael},
LANGUAGE = {eng},
URL = {https://openreview.net/forum?id=LSSiDy7fG1; https://openreview.net/group?id=NeurIPS.cc/2023/Workshop/UniReps},
PUBLISHER = {OpenReview.net},
YEAR = {2023},
MARGINALMARK = {$\bullet$},
BOOKTITLE = {UniReps: The First Workshop on Unifying Representations in Neural Models},
PAGES = {1--6},
ADDRESS = {New Orleans, LA, USA},
}

Endnote

%0 Conference Proceedings
%A Schneider, Jan Philipp
%A Mishal, Fatima
%A Lukasik, Jovita
%A Kolb, Andreas 
%A Keuper, Margret
%A Moeller, Michael
%+ External Organizations
External Organizations
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
External Organizations
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
External Organizations
%T Implicit Representations for Image Segmentation : 
%G eng
%U http://hdl.handle.net/21.11116/0000-000F-240B-D
%U https://openreview.net/forum?id=LSSiDy7fG1
%D 2022
%B First Workshop on Unifying Representations in Neural Models
%Z date of event: 2023-12-15 - 2023-12-15
%C New Orleans, LA, USA
%B UniReps: The First Workshop on Unifying Representations in Neural Models
%P 1 - 6
%I OpenReview.net

2022

Conference paper

J. Lukasik, S. Jung, and M. Keuper

“Learning Where To Look - Generative NAS is Surprisingly Efficient,” in Computer Vision -- ECCV 2022, Tel Aviv, Israel, 2022.

@inproceedings{Lukasik_ECCV2022,
TITLE = {Learning Where To Look -- Generative {NAS} is Surprisingly Efficient},
AUTHOR = {Lukasik, Jovita and Jung, Steffen and Keuper, Margret},
LANGUAGE = {eng},
ISBN = {978-3-031-20049-6},
DOI = {10.1007/978-3-031-20050-2_16},
PUBLISHER = {Springer},
YEAR = {2022},
MARGINALMARK = {$\bullet$},
DATE = {2022},
BOOKTITLE = {Computer Vision -- ECCV 2022},
EDITOR = {Avidan, Shai and Brostow, Gabriel and Ciss{\'e}, Moustapha and Farinella, Giovanni and Hassner, Tal},
PAGES = {257--273},
SERIES = {Lecture Notes in Computer Science},
VOLUME = {13683},
ADDRESS = {Tel Aviv, Israel},
}

Endnote

%0 Conference Proceedings
%A Lukasik, Jovita
%A Jung, Steffen
%A Keuper, Margret
%+ Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
Computer Vision and Machine Learning, MPI for Informatics, Max Planck Society
%T Learning Where To Look - Generative NAS is Surprisingly Efficient : 
%G eng
%U http://hdl.handle.net/21.11116/0000-000A-C00C-0
%R 10.1007/978-3-031-20050-2_16
%D 2022
%B 17th European Conference on Computer Vision
%Z date of event: 2022-10-23 - 2022-10-27
%C Tel Aviv, Israel
%B Computer Vision -- ECCV 2022
%E Avidan, Shai; Brostow, Gabriel; Ciss&#233;, Moustapha; Farinella, Giovanni; Hassner, Tal
%P 257 - 273
%I Springer
%@ 978-3-031-20049-6
%B Lecture Notes in Computer Science
%N 13683
%U https://github.com/jovitalukasik/AG-Net