Ahmed Salem

Ahmed Salem

Postdoc at Microsoft Research

Hello, I’m Ahmed Salem, currently a researcher at the Microsoft Security Response Center (MSRC). Previously, I was a Postdoc researcher at Azure Research after earning my Ph.D. under the guidance of Michael Backes at CISPA, Saarland University.

My research interests are mainly: Machine learning privacy, biomedical data’s privacy and applied cryptography.

What’s New

  • Our paper "Bayesian Estimation of Differential Privacy" got accepted in ICML 2023
  • Our paper "Analyzing Leakage of Personally Identifiable Information in Language Models" got accepted in Oakland 2023
  • Our paper "SoK: Let the Privacy Games Begin! A Unified Treatment of Data Inference Privacy in Machine Learning" got accepted in Oakland 2023
  • Our paper "Two-in-One: A Model Hijacking Attack Against Text Generation Models" got accepted in USENIX Security 2023
  • Our paper "UnGANable: Defending Against GAN-based Face Manipulation" got accepted in USENIX Security 2023
  • I started a PostDoc at Microsoft Research!
  • Our paper "Get a Model! Model Hijacking Attack Against Machine Learning Models" got accepted in NDSS 2022
  • Our paper "ML-Doctor: Holistic Risk Assessment of Inference Attacks Against Machine Learning Models " got accepted in USENIX Security 2022
  • Our paper "BadNL: Backdoor Attacks against NLP Models with Semantic-preserving Improvements" got accepted in ACSAC 2021
  • I will spend 3 months as a research intern at Microsoft Research (MSR) Cambridge!
  • Our technical report titled "Dynamic Backdoor Attacks Against Machine Learning Models" is now online
  • Our paper "Updates-Leak: Data Set Inference and Reconstruction Attacks in Online Learning" got accepted in USENIX Security 2020!
  • Our paper "MemGuard: Defending against Black-Box Membership Inference Attacks via Adversarial Examples" got accepted in CCS 2019!
  • Our technical report titled "Updates-Leak: Data Set Inference and Reconstruction Attacks in Online Learning" is now online
  • Our paper "ML-Leaks: Model and Data Independent Membership Inference Attacks and Defenses on Machine Learning Models " got accepted in NDSS 2019!
  • Our paper titled "Privacy-preserving Similar Patient Queries for Combined Biomedical Data" got accepted in PoPETs 2019
  • Our technical report titled "MLCapsule: Guarded Offline Deployment of Machine Learning as a Service" is now online
  • Our new technical report on membership inference against machine learning models is now online, a previous version of it was presented in PiMLAI 2018