Awesome Adversarial Machine Learning [](https://awesome.re)

A curated list of awesome Machine Learning Security resources.

Also see awesome-ml-for-cybersecurity and The Definitive Security Data Science and Machine Learning Guide.

• Awesome Adversarial Machine Learning
  • Terminology
  • Threat Modeling
  • Controls Guidelines
  • Case Studies
  • Attacks based on domain
  • Attacks based on strategy
  • CTF and Hackathons
  • Frameworks

Terminology

• NIST: A Taxonomy and Terminology of Adversarial Machine Learning

Threat Modeling

• ENISA: Artificial Intelligence Cybersecurity Challenges
• MITRE: Adversarial Threat Landscape for Artificial-Intelligence Systems
• The Threat of Offensive AI to Organizations
• Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey

Controls Guidelines

• ENISA: Securing Machine Learning Algorithms
• AISecOps

Case Studies

• MITRE reports on in-the-wild
• Avito fights content theft using adversarial attacks
• Project Nightshade by researchers from University of Chicago which helps digital artists to protect their works from being used as training data. The key attack method is poisoning.
• Project Glaze by researchers from University of Chicago, similar to Nightshade, but works by mimicry attacks

Attacks based on domain

• Computer Vision
• Speech Recognition
  • Model-specific research
    • Kaldi
    • Lingvo
    • Deepspeech
  • Approaches
    • Man-in-the-Elevator
  • Noise hiding techniques
    • DolphinAttack
    • MPEG Compression

Attacks based on strategy

• Information gathering
  • Membership inference
  • Deanonymization
  • Model inversion
  • Model stealing
  • Blind-spot detection
  • State prediction
• Denial of Service
  • Poisoning DoS
  • Sponge examples
• Biometric Spoofing
  • Master fingerprint
  • Face recognition evasion

CTF and Hackathons

• NIPS 2017: Defense Against Adversarial Attack
• NIPS 2018 : Adversarial Vision Challenge
• GeekPwn CAAD 2018.
• IJCAI-19 Alibaba Adversarial AI Challenge
• GeekPwn CAAD 2019
• Positive Hack Days 2019: AI CTF
• Positive Hack Days 2021: AI CTF
• Positive Hack Days 2022: AI CTF
• UTCTF 2019 (FaceSafe, Bot Protection IV tasks)
• vishwaCTF21 (Good Driver Bad Driver task)
• AI/LLM Exploitation Challenges (AI CTF Labs)

Frameworks

• adversarial-robustness-toolbox
• foolbox
• cleverhans