---
_id: '67300'
abstract:
- lang: eng
  text: Deep neural networks (DNNs) are vulnerable to small adversarial perturbations,
    which are tiny changes to the input data that appear insignificant but cause the
    model to produce drastically different outputs. Many defense methods require modifying
    model architectures during evaluation or performing test-time data purification.
    This not only introduces additional complexity but is often architecture-dependent.
    We show, however, that robust feature learning during training can significantly
    enhance DNN robustness. We propose MOREL, a multi-objective approach that aligns
    natural and adversarial features using cosine similarity and multi-positive contrastive
    losses to encourage similar features for same-class inputs. Extensive experiments
    demonstrate that MOREL significantly improves robustness against both white-box
    and black-box attacks. Our code is available at https://github.com/salomonhotegni/MOREL.
author:
- first_name: Sedjro Salomon
  full_name: Hotegni, Sedjro Salomon
  id: '97995'
  last_name: Hotegni
- first_name: Sebastian
  full_name: Peitz, Sebastian
  id: '47427'
  last_name: Peitz
  orcid: 0000-0002-3389-793X
citation:
  ama: 'Hotegni SS, Peitz S. Enhancing Adversarial Robustness Through Multi-objective
    Representation Learning. In: Senn W, Sanguineti M, Saudargiene A, et al., eds.
    <i>Artificial Neural Networks and Machine Learning – ICANN 2025</i>. Springer
    Nature Switzerland; 2026:442–454. doi:<a href="https://doi.org/10.1007/978-3-032-04558-4_35">10.1007/978-3-032-04558-4_35</a>'
  apa: Hotegni, S. S., &#38; Peitz, S. (2026). Enhancing Adversarial Robustness Through
    Multi-objective Representation Learning. In W. Senn, M. Sanguineti, A. Saudargiene,
    I. V. Tetko, A. E. P. Villa, V. Jirsa, &#38; Y. Bengio (Eds.), <i>Artificial Neural
    Networks and Machine Learning – ICANN 2025</i> (pp. 442–454). Springer Nature
    Switzerland. <a href="https://doi.org/10.1007/978-3-032-04558-4_35">https://doi.org/10.1007/978-3-032-04558-4_35</a>
  bibtex: '@inproceedings{Hotegni_Peitz_2026, place={Cham}, title={Enhancing Adversarial
    Robustness Through Multi-objective Representation Learning}, DOI={<a href="https://doi.org/10.1007/978-3-032-04558-4_35">10.1007/978-3-032-04558-4_35</a>},
    booktitle={Artificial Neural Networks and Machine Learning – ICANN 2025}, publisher={Springer
    Nature Switzerland}, author={Hotegni, Sedjro Salomon and Peitz, Sebastian}, editor={Senn,
    Walter and Sanguineti, Marcello and Saudargiene, Ausra and Tetko, Igor V. and
    Villa, Alessandro E. P. and Jirsa, Viktor and Bengio, Yoshua}, year={2026}, pages={442–454}
    }'
  chicago: 'Hotegni, Sedjro Salomon, and Sebastian Peitz. “Enhancing Adversarial Robustness
    Through Multi-Objective Representation Learning.” In <i>Artificial Neural Networks
    and Machine Learning – ICANN 2025</i>, edited by Walter Senn, Marcello Sanguineti,
    Ausra Saudargiene, Igor V. Tetko, Alessandro E. P. Villa, Viktor Jirsa, and Yoshua
    Bengio, 442–454. Cham: Springer Nature Switzerland, 2026. <a href="https://doi.org/10.1007/978-3-032-04558-4_35">https://doi.org/10.1007/978-3-032-04558-4_35</a>.'
  ieee: 'S. S. Hotegni and S. Peitz, “Enhancing Adversarial Robustness Through Multi-objective
    Representation Learning,” in <i>Artificial Neural Networks and Machine Learning
    – ICANN 2025</i>, 2026, pp. 442–454, doi: <a href="https://doi.org/10.1007/978-3-032-04558-4_35">10.1007/978-3-032-04558-4_35</a>.'
  mla: Hotegni, Sedjro Salomon, and Sebastian Peitz. “Enhancing Adversarial Robustness
    Through Multi-Objective Representation Learning.” <i>Artificial Neural Networks
    and Machine Learning – ICANN 2025</i>, edited by Walter Senn et al., Springer
    Nature Switzerland, 2026, pp. 442–454, doi:<a href="https://doi.org/10.1007/978-3-032-04558-4_35">10.1007/978-3-032-04558-4_35</a>.
  short: 'S.S. Hotegni, S. Peitz, in: W. Senn, M. Sanguineti, A. Saudargiene, I.V.
    Tetko, A.E.P. Villa, V. Jirsa, Y. Bengio (Eds.), Artificial Neural Networks and
    Machine Learning – ICANN 2025, Springer Nature Switzerland, Cham, 2026, pp. 442–454.'
date_created: 2026-10-01T11:53:11Z
date_updated: 2026-10-01T11:53:46Z
department:
- _id: '655'
doi: 10.1007/978-3-032-04558-4_35
editor:
- first_name: Walter
  full_name: Senn, Walter
  last_name: Senn
- first_name: Marcello
  full_name: Sanguineti, Marcello
  last_name: Sanguineti
- first_name: Ausra
  full_name: Saudargiene, Ausra
  last_name: Saudargiene
- first_name: Igor V.
  full_name: Tetko, Igor V.
  last_name: Tetko
- first_name: Alessandro E. P.
  full_name: Villa, Alessandro E. P.
  last_name: Villa
- first_name: Viktor
  full_name: Jirsa, Viktor
  last_name: Jirsa
- first_name: Yoshua
  full_name: Bengio, Yoshua
  last_name: Bengio
keyword:
- own
- own-conference
language:
- iso: eng
page: 442–454
place: Cham
publication: Artificial Neural Networks and Machine Learning – ICANN 2025
publication_identifier:
  isbn:
  - 978-3-032-04558-4
publisher: Springer Nature Switzerland
status: public
title: Enhancing Adversarial Robustness Through Multi-objective Representation Learning
type: conference
user_id: '47427'
year: '2026'
...