Deep Backdoors in Deep Reinforcement Learning Agents

Vasilios MavroudisJamie GawithSañyam VyasChris Hicks

Red TeamBlack Hat USA31 minExpert Open on YouTube ↗

Presenters detail vulnerability discovery and exploitation in Deep Backdoors in Deep Reinforcement Learning Agents.

Summary

Research on planting deep backdoors in deep RL agents.

Related talks

Black Hat USA·40 min

Black Hat
Red Team
Date: 2024
Location: Black Hat

Reinforcement Learning for Autonomous Resilient Cyber Defense

Sara Farmer · David Foster · James Short · Dr Andy Harrison FORS · Ian Miles · Matthew Lacey · Chris Willis · Daniel Harrold · Chris Parry · Gregory Palmer · John H · Tom Wilson · Ryan Heartfield · Mark Dorn · Demian Till · David Rimmer · Samuel Bailey · Peter Haubrick · Jack Stone · Madeline Cheah · Pedro Marques · Alfie Beard · Jonathan Francis Roscoe · Alec Wilson · Ryan Menzies · Marco Casassa Mont · Neela Morarji · Lisa Gralewski · Esin Turkbeyler

Industry experts analyze emerging threats surrounding Reinforcement Learning for Autonomous Resilient Cyber Defense.

cyber-defensereinforcement-learningreinforcement+3

CONFidence·35 min

CONFIDENCE
Red Team
Date: 2024
Location: Confidence

Jakub Strykowski: Reinforcement Learning może zarówno grać w DOOMa, jak i atakować sieci

Jakub Strykowski

Engineers explore architecture and security mechanisms for Jakub Strykowski: Reinforcement Learning może zarówno grać w DOOMa, jak i atakować sieci.

reinforcement-learningjakubstrykowski+3

Strange Loop Conference·30 min

STRANGE LOOP
Red Team
Date: 2022
Location: Strange Loop

"Realtime Hybrid Reinforcement Learning at Scale" by Kexin Xie and Yuxi Zhang

Presenters detail vulnerability discovery and exploitation in "Realtime Hybrid Reinforcement Learning at Scale" by Kexin Xie and Yuxi Zhang.

reinforcement-learningrealtimehybrid+3