mlsanityData to 14 Oct 2026

POLAR: Policy-based Layerwise Reinforcement Learning Method for Stealthy Backdoor Attacks in Federated Learning

arXiv 2510.19056 ↗First posted 21 Oct 2025cs.LGcited by 0

On the map: Deep reinforcement learning methods → Adversarial robustness and RL in finance

Abstract (arXiv)

Federated Learning (FL) enables decentralized model training across multiple clients without exposing local data, but its distributed feature makes it vulnerable to backdoor attacks. Despite early FL backdoor attacks modifying entire models, recent studies have explored the concept of backdoor-critical (BC) layers, which poison the chosen influential layers to maintain stealthiness while achieving high effectiveness. However, existing BC layers approaches rely on rule-based selection without consideration of the interrelations between layers, making them ineffective and prone to detection by advanced defenses. In this paper, we propose POLAR (POlicy-based LAyerwise Reinforcement learning), the first pipeline to creatively adopt RL to solve the BC layer selection problem in layer-wise backdoor attack. Different from other commonly used RL paradigm, POLAR is lightweight with Bernoulli sampling. POLAR dynamically learns an attack strategy, optimizing layer selection using policy gradient updates based on backdoor success rate (BSR) improvements. To ensure stealthiness, we introduce a regularization constraint that limits the number of modified layers by penalizing large attack footprints. Extensive experiments demonstrate that POLAR outperforms the latest attack methods by up to 40% against six state-of-the-art (SOTA) defenses.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Adversarial robustness and RL in finance, one of 15 directions in Deep reinforcement learning methods. Builds on 16 of its 29 references that are in the map. First posted on arXiv 21 Oct 2025.

Builds on

16 of 29 references.

References · 16

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  3. 2016Communication-Efficient Learning of Deep Networks from Decentralized Datacited by 7,669
  4. 2016Pruning Filters for Efficient ConvNetscited by 1,143
  5. 2018Efficient Neural Architecture Search via Parameter Sharingcited by 837
  6. 2018How To Backdoor Federated Learningcited by 765
  7. 2018Local SGD Converges Fast and Communicates Littlecited by 605
  8. 2019Local Model Poisoning Attacks to Byzantine-Robust Federated Learningcited by 493
  9. 2018Analyzing Federated Learning through an Adversarial Lenscited by 351
  10. 2020FLTrust: Byzantine-robust Federated Learning via Trust Bootstrappingcited by 346
  11. 2020FetchSGD: Communication-Efficient Federated Learning with Sketchingcited by 179
  12. 2022FLDetector: Defending Federated Learning Against Model Poisoning Attacks via Detecting Malicious Clientscited by 125
4 more
  1. 2020Defending against Backdoors in Federated Learning with Robust Learning Ratecited by 82
  2. 2023Backdoor Federated Learning by Poisoning Backdoor-Critical Layerscited by 18
  3. 2023Learning to Backdoor Federated Learningcited by 8
  4. 2024SDBA: A Stealthy and Long-Lasting Durable Backdoor Attack in Federated Learningcited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Enhancing the Effectiveness and Durability of Backdoor Attacks in Federated Learning through Maximizing Task Distinction
  2. 2026FedSurrogate: Backdoor Defense in Federated Learning via Layer Criticality and Surrogate Replacement
  3. 2026BackDFL: A Unified Benchmark For Backdoor Attacks and Defenses In Decentralized Federated Learning
  4. 2025SPA: Towards More Stealth and Persistent Backdoor Attacks in Federated Learning
  5. 2025FedBAP: Backdoor Defense via Benign Adversarial Perturbation in Federated Learning
  6. 2025DOPA: Stealthy and Generalizable Backdoor Attacks from a Single Client under Challenging Federated Constraints ↗
  7. 2023FedGrad: Mitigating Backdoor Attacks in Federated Learning Through Local Ultimate Gradients Inspection
  8. 2025TrojanDam: Detection-Free Backdoor Defense in Federated Learning through Proactive Model Robustification utilizing OOD Data