Institution profile

TU Wien

Academic institutioneurope · at
Official website
Research library659linked papers
Opportunities0open roles
Selected work

Representative Papers

The Complexity of Bayesian Network Learning: Revisiting the Superstructure

Feb 10, 2026Neural Information Processing Systems

This study investigates the parameterized complexity of Bayesian network structure learning (BNSL) under super-structure constraints. By integrating graph-theoretic parameters—such as feedback edge set size, local feedback edge set size, and treewidth—with input representation formats, particularly additive representations, the work systematically analyzes the fixed-parameter tractability of BNSL. The main contributions include the first proof that BNSL is fixed-parameter tractable when parameterized by the size of a feedback edge set, a result extended to local feedback edge sets. Furthermore, it establishes that under additive representations, treewidth alone suffices for fixed-parameter tractability—a finding that also applies to Polytree learning. The paper provides a complete complexity classification across mainstream graph parameters and derives corresponding conditional lower bounds, thereby significantly advancing the theoretical foundation of BNSL.

30 citations2 influentialRead paper

LLMs as Hackers: Autonomous Linux Privilege Escalation Attacks

Oct 17, 2023

This work investigates the capabilities and limitations of large language models (LLMs) in automating Linux local privilege escalation during penetration testing. We propose the first end-to-end, LLM-driven autonomous privilege escalation framework, incorporating dynamic error recovery, context-aware memory management, and multi-stage guided reasoning to support both in-context learning and interactive command-line inference. We systematically evaluate GPT-4-turbo, GPT-3.5-turbo, and Llama3 on real Linux target machines. Results show GPT-4-turbo successfully exploits 33–83% of known privilege escalation vulnerabilities—substantially outperforming GPT-3.5-turbo (16–50%) and Llama3 (0–33%). This study bridges a critical gap in AI-powered red teaming by enabling automated lateral movement and privilege escalation, and empirically establishes the feasibility and practical boundaries of advanced reasoning LLMs in ethical hacking tasks.

10 citations3 influentialRead paper

Computing the Schulze Method for Large-Scale Preference Data Sets

Jul 01, 2018International Joint Conference on Artificial Intelligence

To address the poor computational efficiency and scalability of the Schulze method on large-scale preference data, this paper proposes a high-performance parallel algorithm built upon the Pregel graph-processing framework. We first prove that Schulze winner determination is NL-complete—strictly below P-complete—thereby establishing its intrinsic parallel tractability and providing a theoretical foundation for efficient distributed implementation. Leveraging a weighted directed graph model over candidates, we design a novel parallel shortest-path optimization strategy, which significantly outperforms P-complete alternatives such as Ranked Pairs. Experimental evaluation demonstrates strong scalability to datasets with up to one million voters and candidates, achieving near-linear speedup on real-world data. The approach thus bridges deep theoretical insight—rooted in complexity-theoretic analysis—with practical engineering utility.

9 citations1 influentialRead paper

CWASI: A WebAssembly Runtime Shim for Inter-Function Communication in the Serverless Edge-Cloud Continuum

Dec 06, 2023IFIP International Information Security Conference

To address high latency and network overhead caused by frequent inter-function data exchange relying on remote storage in edge–cloud serverless continua, this paper proposes Shim, a location-aware WebAssembly runtime. Shim introduces a novel tri-modal communication model—local, proximal, and remote—and is the first OCI-compliant WASI runtime to enable dynamic, adaptive communication-path selection and transparent optimization. By deeply integrating function hosting mechanisms with location-aware routing, Shim achieves up to 95% reduction in inter-function communication latency and a 30× throughput improvement in co-located deployments. These gains significantly alleviate cross-layer network dependency and transmission overhead. The work establishes a new paradigm for efficient data exchange in serverless architectures operating across edge and cloud environments.

7 citationsRead paper
Recent publications

Latest Papers