Lukas Pirch
Scholar

Lukas Pirch

Google Scholar ID: HGIT8psAAAAJ
PhD Candidate, TU Berlin
graph-based machine learningvulnerability discovery
Homepage↗Google Scholar↗
Citations & Impact
All-time
Citations
352
 
H-index
5
 
i10-index
5
 
Publications
10
 
Co-authors
0
 
Contact
Emaillukas.pirch@tu-berlin.deGitHubOpen ↗
Publications
3 items
LLM-based Vulnerability Discovery through the Lens of Code Metrics
2025
Cited
0
SoK: Where to Fuzz? Assessing Target Selection Methods in Directed Fuzzing
ACM Asia Conference on Computer and Communications Security · 2024
Cited
1
Manipulating Feature Visualizations with Gradient Slingshots
arXiv.org · 2024
Cited
6
Resume (English only)
Academic Achievements
  • Paper accepted at ICSE 2026: 'LLM-based Vulnerability Discovery through the Lens of Code Metrics' (to appear)
  • Paper accepted at NeurIPS 2025: 'Manipulating Feature Visualizations with Gradient Slingshots' (to appear)
  • Published at ICML 2025: 'Adversarial Inputs for Linear Algebra Backends'
  • Two papers published at ASIACCS 2024: 'SoK: Where to Fuzz? Assessing Target Selection Methods in Directed Fuzzing' and 'Cross-Language Differential Testing of JSON Parsers'
  • Published at AISEC 2023: 'Broken Promises: Measuring Confounding Effects in Learning-based Vulnerability Discovery'
  • Published at NDSS 2023: 'Machine Unlearning of Features and Labels'
  • Published at EuroSec 2021: 'TagVet: Vetting Malware Tags using Explainable Machine Learning'
  • Technical report (2020): 'Against All Odds: Winning the Defense Challenge in an Evasion Competition with Diversification' (arXiv:2010.09569)
Co-authors
0 total
Co-authors: 0 (list not available)