Hasso-Plattner-Institut
Prof. Dr. Tobias Friedrich
 

27.03.2023

Two Papers accepted at PAKDD and ACNS

We are proud to announce two recently accepted papers written by our group members. First, Philipp Fischbeck and others authored the paper The Common-Neighbors Metric is Noise-Robust & Reveals Substructures of Real-World Networks for the Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD) in Osaka, Japan on 25-28 May. The problem considered by the paper is the following: Given a structured network with randomly inserted edges (simulating noise/outliers), can we recognize these outliers? The authors use the simple common-neighbors metric. They provide both mathematically rigorous analyses on special random-graph models as well as empirical analyses on real-world networks. Their results show that the metric is a very good classifier that can reliably detect noise up to extreme levels.

For the International Conference on Applied Cryptography and Network Security (ACNS) on 19-22 June in Kyoto, Japan, the paper Analysis and Prevention of Averaging Attacks against Obfuscation Protocols was written by Gregor Lagodzinski with co-authors at SAP. The paper studies the risk of using independent randomness in obfuscating techniques used in shared data scenarios like a cloud. In particular, it is shown that averaging attacks reveal information exponentially fast. In the second part, the paradigm of data dependent deterministic obfuscation (D^3O) is proposed, which uses deterministic randomness, and the effectiveness and negligible runtime overhead are illustrated.

  • The Common-Neighbors Metr... - Download
    Cohen, Sarel; Fischbeck, Philipp; Friedrich, Tobias; Krejca, Martin S. The Common-Neighbors Metric is Noise-Robust and Reveals Substructures of Real-World NetworksPacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD) 2023: 67–79
     
  • Analysis and Prevention o... - Download
    Becher, Kilian; Lagodzinski, J. A. Gregor; Parra-Arnau, Javier; Strufe, Thorsten Analysis and Prevention of Averaging Attacks Against Obfuscation ProtocolsApplied Cryptography and Network Security (ACNS), Part {I} 2023: 451–475