Lune

CVPR Findings2026

TriGuard-FL: A User-Centric Trust Triad in Federated Learning via Auditable Data, Verifiable Contributions, and Antidote-Driven Mitigation

K Naveen Kumar, Mohsen Guizani

Abstract

Trust in federated learning (FL) extends beyond robustness as it demands transparency, accountability, and proactive resilience against adversarial clients. Despite recent FL progress, achieving this trust triad remains an open and challenging research problem, as existing methods primarily rely on server-centric Byzantine aggregation. This limits insight into local data quality and overlooks distributionlevel valuation, which captures data heterogeneity. However, hard rejection of anomalous updates before aggregation weakens global model generalization, especially under non-IID & cross-silo FL settings with limited clients. Hence, we propose TriGuard-FL, a novel 'user-centric' trust triad that enables auditable distribution-level data valuation, verifiable contribution reliability, and adaptive mitigation of anomalous updates through refinement rather than rejection. In TriGuard-FL, (1) clients conduct latentspace Mahalanobis data valuation and (2) compute maximum mean discrepancy based trust scores from privacyprotected distributional statistic moments shared with the server. Using these scores, the server applies the proposed antidote gradient injection (AGI) strategy to adaptively neutralize malicious updates rather than discarding them through trust-scaled gradient modulation. Our theoretical analysis demonstrates convergence, computational efficiency, auditability, verifiability, and mitigability of TriGuard-FL against adversarial clients. Experiments on four benchmarks, including medical vision and financial tabular datasets, and three adversarial attacks across crossdevice & cross-silo FL settings, show that TriGuard-FL reduces accuracy degradation under attack while preserving clean accuracy and efficiency, establishing a trust triad in FL from an end-user perspective.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

Questions to start from

Lune can draw on

the full paper · related work

Get the answer

Free to start. No credit card required.

Lineage

Builds onCited byLeft to right by year; the more cited, the higher.

Builds on21

  1. Adaptive Federated OptimizationSashank J. Reddi, Zachary Charles, Manzil Zaheer, Zachary Garrett et al.ICLR 2021 · 1,917 citations
  2. BatchCrypt: Efficient Homomorphic Encryption for Cross-Silo Federated LearningChengliang Zhang, Suyi Li, Junzhe Xia, Wei Wang et al.USENIX ATC 2020 · 967 citations
  3. Back to the Drawing Board: A Critical Evaluation of Poisoning Attacks on Production Federated LearningVirat Shejwalkar, Amir Houmansadr, Peter Kairouz, Daniel RamageS&P 2022 · 302 citations
  4. Diverse Client Selection for Federated Learning via Submodular MaximizationRavikumar Balakrishnan, Tian Li, Tianyi Zhou, Nageen Himayat et al.ICLR 2022 · 140 citations
  5. FLShield: A Validation Based Federated Learning Framework to Defend Against Poisoning AttacksEhsanul Kabir, Zeyu Song, Md. Rafi Ur Rashid, Shagufta MehnazS&P 2024 · 32 citations

Related papers

  1. RECESS Vaccine for Federated Learning: Proactive Defense Against Model Poisoning AttacksHaonan Yan, Wenjing Zhang, Qian Chen, Xiaoguang Li et al.NeurIPS 2023 · 26 citations
  2. FLTrust: Byzantine-robust Federated Learning via Trust BootstrappingXiaoyu Cao, Minghong Fang, Jia Liu, Neil Zhenqiang GongNDSS 2021
  3. SAFER-FL: Adversarially Robust Federated Learning in IoT Networks via Latent Space Auditing and Verifiable ContributionsNaveen Kumar Kummari, Mohsen GuizaniINFOCOM 2026
  4. Bant: Byzantine Antidote via Trial Function and Trust ScoresGleb Molodtsov, Daniil Medyakov, Sergey Skorik, Nikolas Khachaturov et al.AAAI 2026 · 4 citations
  5. Do We Really Need to Design New Byzantine-robust Aggregation Rules?Minghong Fang, Seyedsina Nabavirazavi, Zhuqing Liu, Wei Sun et al.NDSS 2025
Dusk over the sea between two cliffs drawn in fine vertical lines