An analytical attack that enables a malicious server to recover private user data by leveraging a poisoned pretrained model and fine-tuning parameters is proposed, and results on both language and vision tasks demonstrate high-fidelity data recovery across multiple baselines.
Abstract
Parameter-efficient fine-tuning (PEFT), such as low-rank adaptation (LoRA), has recently been adopted in federated learning to reduce communication and computation costs. In this setup, users download a pretrained model from the server prior to fine-tuning, and then fine-tune lightweight LoRA modules locally while keeping the pretrained model frozen, sharing only the gradients of the fine-tuning parameters with the server. Despite its growing popularity, robustness of federated fine-tuning against an adversarial server remains underexplored, where the server maliciously tampers with the training protocol to breach the privacy of users'data. In this work, we investigate gradient inversion attacks on LoRA fine-tuning. We propose an analytical attack that enables a malicious server to recover private user data by leveraging a poisoned pretrained model and fine-tuning parameters. Our design embeds fine-tuning data within the shared gradients, to allow the server to analytically reconstruct user data. Unlike prior works, our attack is applicable to both language and vision tasks, does not rely on computationally expensive (adversarial) pretraining with public datasets or require the number of training tokens to be less than the rank of LoRA modules. Experimental results on both language and vision tasks demonstrate high-fidelity data recovery across multiple baselines, revealing several critical vulnerabilities.
Federated fine-tuning is bottlenecked by communication: FedAvg and pseudo-gradient schemes transmit a payload that scales with the model, and gradient compression shrinks it by only a constant factor. We take a different lever. Mapping networks generate a network's weights from a small trainable latent through a frozen affine projection; because the map is shared and affine, averaging latents is exactly averaging the generated weights. We turn this into a practical low-bandwidth federated channel with two changes: a low-rank, seed-regenerable factorisation of the projection (cutting generator memory from ~80 GB to ~10 MB), and a delta formulation $\theta = \theta^{\mathrm{pre}} + U V^{\top} z$ that learns an additive correction around a shared centrally-pretrained base -- federated fine-tuning, which is what makes the method work at scale. A frozen orthogonal classifier head further removes the head from the payload while improving accuracy. On CIFAR-100 with ResNet-18+GroupNorm, our method (FLITE, Federated Low-rank Iterative Training Engine) communicates 1,280 floats (~5 KB) per client per round -- an 8718x reduction -- and reaches 74.67%, within ~0.5 pp of full-weight FedAvg. The averaging identity holds to floating-point precision ($6 \times 10^{-8}$); the method sits one to two orders of magnitude below PowerSGD and top-k on the bandwidth-accuracy Pareto; it matches or exceeds full-weight FedAvg under strong non-IID skew. int4 latents reach 648 bytes per round at unchanged accuracy, whereas int4 full-weight FedAvg collapses to chance.
Federated learning is appealing for privacy-sensitive network systems, yet its practical deployment remains hindered by the following three recurring challenges: (1) client drift under non-IID data, (2) vulnerability to corrupted updates, and (3) the communication cost of repeated model exchange. Most existing approaches address these issues in isolation. While analytically convenient, this separation often fails to reflect real-world conditions. For instance, defenses against poisoning may suppress useful updates, while personalization and compression can alter the aggregation geometry itself. In this paper, we study these effects jointly and propose URP-FL, a compact training framework that integrates reliability-aware aggregation, local regularization for drift control, and sparse client uploads. We provide theoretical analysis establishing a convergence bound with distinct terms capturing optimization error, data heterogeneity, and adversarial impact. Experiments on a non-IID image classification benchmark with sign-flip and label-flip attacks demonstrate the benefits of the unified design. Compared to FedAvg and FedProx, this URP-FL maintains accuracy under attack while reducing transmitted parameters by approximately 75%. Rather than presenting a production ready system, it offers a reproducible and technically coherent step toward federated learning that is more robust under realistic conditions.
Hua Kun, Wei Wang· 2026 International Conferenc...· 0 citations
TRISHUL is proposed, a spectral-control framework for robust federated PEFT that follows the FL no-raw-data-sharing setting but does not itself provide formal privacy guarantees, and improves convergence, stability, and final performance over federated LoRA baselines.
This work proposes BMAT (Bilevel-Minimax Adversarial Transfer), an integrated bottom-up solver that combines a Soft Weight Modulator and an Implicit Gradient Approximator to enable ternary coupling among initialization, surrogate adaptation, and perturbation optimization.
The Krum-Proxy attack is introduced, a selection-aware backdoor injection strategy that consistently bypasses Byzantine-robust aggregation and constructs adversarial updates that are not only similar to benign updates but are also optimized to lie in regions of the update space that are favored during aggregation.
Srinivasan Subramanian, Md Abdullah Al Hafiz Khan, K. A. Islam· 2026 International Conferenc...· 0 citations
SplitLite is proposed, a communication-efficient split federated LoRA fine-tuning method that exploits the low effective rank structure of consecutive-epoch activation and gradient residuals, thereby significantly reducing both activation uplink and gradient downlink traffic.
Tao Li, Yulin Tang, Qi Guo et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
