Backdoors in DRL: Four Environments Focusing on In-distribution Triggers

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ashcraft, Chace, Staley, Ted, Carney, Josh, Hickert, Cameron, Juba, Derek, Karra, Kiran, Drenkow, Nathan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914198933471232
author Ashcraft, Chace
Staley, Ted
Carney, Josh
Hickert, Cameron
Juba, Derek
Karra, Kiran
Drenkow, Nathan
author_facet Ashcraft, Chace
Staley, Ted
Carney, Josh
Hickert, Cameron
Juba, Derek
Karra, Kiran
Drenkow, Nathan
contents Backdoor attacks, or trojans, pose a security risk by concealing undesirable behavior in deep neural network models. Open-source neural networks are downloaded from the internet daily, possibly containing backdoors, and third-party model developers are common. To advance research on backdoor attack mitigation, we develop several trojans for deep reinforcement learning (DRL) agents. We focus on in-distribution triggers, which occur within the agent's natural data distribution, since they pose a more significant security threat than out-of-distribution triggers due to their ease of activation by the attacker during model deployment. We implement backdoor attacks in four reinforcement learning (RL) environments: LavaWorld, Randomized LavaWorld, Colorful Memory, and Modified Safety Gymnasium. We train various models, both clean and backdoored, to characterize these attacks. We find that in-distribution triggers can require additional effort to implement and be more challenging for models to learn, but are nevertheless viable threats in DRL even using basic data poisoning attacks.
format Preprint
id arxiv_https___arxiv_org_abs_2505_17248
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Backdoors in DRL: Four Environments Focusing on In-distribution Triggers
Ashcraft, Chace
Staley, Ted
Carney, Josh
Hickert, Cameron
Juba, Derek
Karra, Kiran
Drenkow, Nathan
Machine Learning
Cryptography and Security
Backdoor attacks, or trojans, pose a security risk by concealing undesirable behavior in deep neural network models. Open-source neural networks are downloaded from the internet daily, possibly containing backdoors, and third-party model developers are common. To advance research on backdoor attack mitigation, we develop several trojans for deep reinforcement learning (DRL) agents. We focus on in-distribution triggers, which occur within the agent's natural data distribution, since they pose a more significant security threat than out-of-distribution triggers due to their ease of activation by the attacker during model deployment. We implement backdoor attacks in four reinforcement learning (RL) environments: LavaWorld, Randomized LavaWorld, Colorful Memory, and Modified Safety Gymnasium. We train various models, both clean and backdoored, to characterize these attacks. We find that in-distribution triggers can require additional effort to implement and be more challenging for models to learn, but are nevertheless viable threats in DRL even using basic data poisoning attacks.
title Backdoors in DRL: Four Environments Focusing on In-distribution Triggers
topic Machine Learning
Cryptography and Security
url https://arxiv.org/abs/2505.17248