Portrait

Lukas Pirch

Room
B 118
Address
FG Machine Learning and Security
Technische Universität Berlin
Ernst-Reuter-Platz 7
10587 Berlin, Germany

About me

I am a PhD student at Technische Universität Berlin. I am part of the Chair of Machine Learning and Security within the Berlin Institute for the Foundations of Learning and Data (BIFOLD).

Publications

SoK: Where to Fuzz? Assessing Target Selection Methods in Directed Fuzzing.
Felix Weißberg, Jonas Möller, Tom Ganz, Erik Imgrund, Lukas Pirch, Lukas Seidel, Moritz Schloegel, Thorsten Eisenhofer and Konrad Rieck.
Proc. of the 19th ACM Asia Conference on Computer and Communications Security (ASIACCS), 2024.

PDF Code Data

Cross-Language Differential Testing of JSON Parsers.
Jonas Möller, Felix Weißberg, Lukas Pirch, Thorsten Eisenhofer and Konrad Rieck.
Proc. of the 19th ACM Asia Conference on Computer and Communications Security (ASIACCS), 2024.

PDF Code

Manipulating Feature Visualizations with Gradient Slingshots.
Dilyara Bareeva, Marina Höhne, Alexander Warnecke, Lukas Pirch, Klaus-Robert Müller, Konrad Rieck and Kirill Bykov.
Technical report, arXiv:2401.06122, 2024.

Link

Broken Promises: Measuring Confounding Effects in Learning-based Vulnerability Discovery.
Erik Imgrund, Tom Ganz, Martin Härterich, Niklas Risse, Lukas Pirch and Konrad Rieck.
Proc. of the 16th ACM Workshop on Artificial Intelligence and Security (AISEC), 2023.

PDF

Machine Unlearning of Features and Labels.
Alexander Warnecke, Lukas Pirch, Christian Wressnegger and Konrad Rieck.
Proc. of the 30th Network and Distributed System Security Symposium (NDSS), 2023.

PDF Code

TagVet: Vetting Malware Tags using Explainable Machine Learning.
Lukas Pirch, Alexander Warnecke, Christian Wressnegger and Konrad Rieck.
Proc. of the 14th ACM European Workshop on Systems Security (EuroSec), 2021.

PDF

Against All Odds: Winning the Defense Challenge in an Evasion Competition with Diversification.
Erwin Quiring, Lukas Pirch, Michael Reimsbach, Daniel Arp and Konrad Rieck.
Technical report, arXiv:2010.09569, 2020.

PDF Link Code