Institution profile

CISPA Helmholtz Center for Information Security

Academic institutioneurope · de
Official website
Research library443linked papers
Opportunities0open roles
Selected work

Representative Papers

Fake It Until You Break It: On the Adversarial Robustness of AI-generated Image Detectors

Oct 02, 2024arXiv.org

This study addresses the insufficient adversarial robustness of AI-generated image detectors in real-world settings, where they are vulnerable to black-box attacks and common social media degradations (e.g., JPEG compression, resizing, color distortion), enabling malicious misuse for disinformation and undermining democratic trust. We conduct the first systematic evaluation of mainstream detectors under combined black-box adversarial perturbations and realistic degradations, revealing that state-of-the-art models suffer over 40% accuracy degradation without model access. To mitigate this, we propose a lightweight CLIP-enhanced defense grounded in zero-shot detection and black-box transfer attack modeling—requiring no retraining or fine-tuning. Our method preserves original detection performance while reducing adversarial success rates by 76%, substantially restoring practical utility and robustness on real platforms. This work delivers a deployable, trustworthy solution for AI-generated content authentication.

3 citations1 influentialRead paper

Logics and Algorithms for Hyperproperties

Apr 01, 2023ACM SIGLOG News

This paper addresses hyperproperties—higher-order system requirements encompassing information-flow security, knowledge reasoning, and robustness, which span multiple execution traces—by proposing the first unified logical and algorithmic framework covering the entire verification lifecycle. Methodologically, it rigorously characterizes the expressive power and decidability boundaries of classical temporal logics (LTL, CTL, S1S) over hyperproperties; then introduces a novel multi-trace synchronization modeling and quantifier alternation handling mechanism grounded in higher-order temporal logic, constraint solving, and symbolic automata. Key contributions include: (i) a comprehensive taxonomy and complexity-theoretic characterization of hyperproperty logics; (ii) an open-source verification toolchain supporting HyperLTL and HyperCTL*; and (iii) end-to-end support for core verification tasks—including satisfiability checking, model checking, runtime monitoring, and controller synthesis.

2 citationsRead paper

Is a LOCAL algorithm computable?

Feb 24, 2026

This work investigates whether node state updates in the LOCAL model of distributed computing should be restricted to computable functions and how this restriction affects the round complexity of Locally Checkable Labeling (LCL) problems. By integrating techniques from distributed computing theory, computability analysis, and LCL complexity classification, the paper establishes for the first time that the round complexity of LCL problems is jointly determined by the computability assumption and prior knowledge of the graph size $n$. The central contribution is the construction of an LCL problem $\Pi$ that requires $\Omega(\sqrt{n})$ rounds in the computable LOCAL model without knowledge of $n$, yet can be solved in $O(\log n)$ rounds either when an upper bound on $n$ is known or when non-computable updates are permitted. Moreover, the authors demonstrate that this complexity separation is not isolated but exhibits a general phenomenon.

1 citationsRead paper

Fixed Aggregation Features Can Rival GNNs

Jan 27, 2026

This work challenges the prevailing assumption that graph neural networks (GNNs) require trainable neighborhood aggregation to effectively learn node representations. We propose Fixed Aggregation Features (FAFs), a training-free neighborhood aggregation scheme—such as simple averaging—that transforms graph learning tasks into tabular data problems, enabling the use of well-established tabular models like MLPs for efficient and interpretable node representation learning. Grounded in the Kolmogorov–Arnold representation theorem, our theoretical analysis supports this approach. Empirical results across 14 benchmark datasets show that FAF+MLP matches or exceeds the performance of state-of-the-art GNNs and graph Transformers on 12 datasets, underperforming only slightly on Roman Empire and Minesweeper—tasks that demand extremely deep architectures—thereby overturning conventional wisdom regarding the necessity of trainable aggregation.

1 citationsRead paper

SL-CBM: Enhancing Concept Bottleneck Models with Semantic Locality for Better Interpretability

Jan 19, 2026

This work addresses the limited spatial locality in existing Concept Bottleneck Models (CBMs), which hinders precise alignment between concepts and semantically meaningful image regions, thereby undermining interpretability credibility. To overcome this limitation, the authors introduce, for the first time in CBMs, a combination of 1×1 convolutions and cross-attention mechanisms to generate faithful saliency maps tightly coupled with the model’s reasoning process. They further employ contrastive and entropy regularization to jointly optimize prediction accuracy, map sparsity, and explanation fidelity. Extensive experiments demonstrate that the proposed approach significantly improves local concept-region alignment, enhances explanation clarity, and boosts intervention effectiveness across multiple image datasets, all while maintaining competitive classification performance.

1 citationsRead paper
Recent publications

Latest Papers