Zobrazeno 1 - 10
of 3 413
pro vyhledávání: '"Gosch, A."'
Existing studies have shown that Graph Neural Networks (GNNs) are vulnerable to adversarial attacks. Even though Graph Transformers (GTs) surpassed Message-Passing GNNs on several benchmarks, their adversarial robustness properties are unexplored. Ho
Externí odkaz:
http://arxiv.org/abs/2407.11764
Generalization of machine learning models can be severely compromised by data poisoning, where adversarial changes are applied to the training data. This vulnerability has led to interest in certifying (i.e., proving) that such changes up to a certai
Externí odkaz:
http://arxiv.org/abs/2407.10867
Most adversarial attacks and defenses focus on perturbations within small $\ell_p$-norm constraints. However, $\ell_p$ threat models cannot capture all relevant semantic-preserving perturbations, and hence, the scope of robustness evaluations is limi
Externí odkaz:
http://arxiv.org/abs/2310.04285
We perform the first adversarial robustness study into Graph Neural Networks (GNNs) that are provably more powerful than traditional Message Passing Neural Networks (MPNNs). In particular, we use adversarial robustness as a tool to uncover a signific
Externí odkaz:
http://arxiv.org/abs/2308.08173
Autor:
Gosch, Lukas, Geisler, Simon, Sturm, Daniel, Charpentier, Bertrand, Zügner, Daniel, Günnemann, Stephan
Despite its success in the image domain, adversarial training did not (yet) stand out as an effective defense for Graph Neural Networks (GNNs) against graph structure perturbations. In the pursuit of fixing adversarial training (1) we show and overco
Externí odkaz:
http://arxiv.org/abs/2306.15427
Many works show that node-level predictions of Graph Neural Networks (GNNs) are unrobust to small, often termed adversarial, changes to the graph structure. However, because manual inspection of a graph is difficult, it is unclear if the studied pert
Externí odkaz:
http://arxiv.org/abs/2305.00851
Autor:
Alexander Thomas, Thomas Battenfeld, Ivana Kraiselburd, Olympia Anastasiou, Ulf Dittmer, Ann-Kathrin Dörr, Adrian Dörr, Carina Elsner, Jule Gosch, Vu Thuy Khanh Le-Trilling, Simon Magin, René Scholtysik, Pelin Yilmaz, Mirko Trilling, Lara Schöler, Johannes Köster, Folker Meyer
Publikováno v:
BMC Genomics, Vol 25, Iss 1, Pp 1-9 (2024)
Abstract Background At a global scale, the SARS-CoV-2 virus did not remain in its initial genotype for a long period of time, with the first global reports of variants of concern (VOCs) in late 2020. Subsequently, genome sequencing has become an indi
Externí odkaz:
https://doaj.org/article/7c88bf5065d14a7caf6ac89a4da96eec
Deep learning models are known to put the privacy of their training data at risk, which poses challenges for their safe and ethical release to the public. Differentially private stochastic gradient descent is the de facto standard for training neural
Externí odkaz:
http://arxiv.org/abs/2301.00738
Autor:
Ann-Kathrin Dörr, Josefa Welling, Adrian Dörr, Jule Gosch, Hannah Möhlen, Ricarda Schmithausen, Jan Kehrmann, Folker Meyer, Ivana Kraiselburd
Publikováno v:
GigaByte (2024)
Background Next-generation sequencing for microbial communities has become a standard technique. However, the computational analysis remains resource-intensive. With declining costs and growing adoption of sequencing-based methods in many fields, val
Externí odkaz:
https://doaj.org/article/6bf7926ace6c4923bb256f83148140b6
Publikováno v:
Ceramics, Vol 6, Iss 4, Pp 2243-2255 (2023)
3D printing of ceramics has started gaining traction in architecture over the past decades. However, many existing paste-based extrusion techniques have not yet been adapted or made feasible in ceramics. A notable example is coextrusion, a common app
Externí odkaz:
https://doaj.org/article/2449553e47414e348f174ccb3020df8b