Edge intelligence facilitates privacy-preserving learning across devices with heterogeneous data and resources. While personalized federated learning balances local requirements with global objectives, conventional approaches based on parameter interaction suffer from unaffordable communication burdens and strict architectural constraints. In contrast, logit-interaction-based architectures offer flexibility but fail to address the compounded challenge of model heterogeneity and multimodal disparity without relying on unrealistic public datasets or transmitting high-dimensional features. To tackle this dilemma, we propose FedMCache, a cache-driven personalized federated learning architecture designed for multimodal edge intelligence. FedMCache reserves a modality-aware knowledge cache on the server, which organizes lightweight logits uploaded by clients using compact semantic hashes. During training, the server retrieves and aggregates modality-consistent teacher knowledge to guide the optimization of heterogeneous student models via personalized distillation. This mechanism enables effective collaboration across clients with incomplete modality subsets while eliminating gradient leakage risks. Empirical experiments on BraTS2021 and NYU-Depth-V2 demonstrate that FedMCache achieves competitive accuracy and narrows the gap to centralized training while substantially reducing communication.

Prompt learning is a parameter-efficient approach for vision-language models, yet its robustness under label noise is less investigated. Visual content contains richer and more reliable semantic information, which remains more robust under label noise. However, the prompt itself is highly susceptible to label noise. Motivated by this intuition, we propose VisPrompt, a lightweight and robust vision-guided prompt learning framework for noisy-label settings. Specifically, we exploit a cross-modal attention mechanism to reversely inject visual semantics into prompt representations. This enables the prompt tokens to selectively aggregate visual information relevant to the current sample, thereby improving robustness by anchoring prompt learning to stable instance-level visual evidence and reducing the influence of noisy supervision. To address the instability caused by using the same way of injecting visual information for all samples, despite differences in the quality of their visual cues, we further introduce a lightweight conditional modulation mechanism to adaptively control the strength of visual information injection, which strikes a more robust balance between text-side semantic priors and image-side instance evidence. The proposed framework effectively suppresses the noise-induced disturbances, reduce instability in prompt updates, and alleviate memorization of mislabeled samples. VisPrompt significantly improves robustness while keeping the pretrained VLM backbone frozen and introducing only a small amount of additional trainable parameters. Extensive experiments under synthetic and real-world label noise demonstrate that VisPrompt generally outperforms existing baselines on seven benchmark datasets and achieves stronger robustness. Our code is publicly available at https://github.com/gezbww/Vis_Prompt.

Heterogeneous device-edge-cloud computing infrastructures have become the backbone of modern telecommunication operators and Wide Area Networks (WANs), providing multi-tier computational support for emerging intelligent applications. With the rapid proliferation of Large Language Model (LLM) services, efficiently coordinating inference tasks and reducing communication burden within these multi-tier network architectures becomes a critical deployment challenge. Current LLM serving paradigms exhibit significant limitations: on-device deployment restricts service to lightweight LLMs due to hardware constraints, while cloud-centric deployment encounters resource congestion and considerable prompt communication overhead during peak periods. Model-cascading inference, though better suited for multi-tier networks, depends on static, manually-tuned thresholds that cannot adapt to dynamic network conditions or varying task complexities. To address these challenges, we propose RecServe, a recursive offloading framework tailored for LLM serving in multi-tier networks. RecServe introduces a task-specific hierarchical confidence evaluation mechanism that guides offloading decisions based on inferred task complexity in progressively scaled LLMs across device, edge, and cloud tiers. To further enable intelligent task routing across tiers, RecServe employs a sliding-window-based dynamic offloading strategy with quantile interpolation, enabling real-time tracking of historical confidence distributions and adaptive offloading threshold adjustments. This design allows inference tasks to be recursively offloaded to higher tiers only when necessary, optimizing heterogeneous resource utilization while reducing cross-tier communication with little compromise on service quality. Theoretical analysis provides distinct conditions under which RecServe is expected to achieve reduced communication burden and computational costs. Experiments on eight datasets demonstrate that RecServe outperforms CasServe in both service quality and communication efficiency, and reduces the communication burden by over 50% compared to centralized cloud-based serving.

Parameter-efficient fine-tuning (PEFT) reduces the compute and memory demands of adapting large language models, yet standard low-rank adapters (e.g., LoRA) can lag full fine-tuning in performance and stability because they restrict updates to a fixed rank-r subspace. We propose Matrix-Transformation based Low-Rank Adaptation (MTLoRA), a brain-inspired extension that inserts a learnable r × r transformation T into the low-rank update (ΔW=BTA). By endowing the subspace with data-adapted geometry (e.g., rotations, scalings, and shears), MTLoRA reparameterizes the rank-r hypothesis class, improving its conditioning and inductive bias at negligible O(r2) overhead, and recovers LoRA when T=Ir. We instantiate four structures for T-SHIM (T=C), ICFM (T=CC⊤), CTCM (T=CD), and DTSM (T=C+D)-providing complementary inductive biases (change of basis, PSD metric, staged mixing, dual superposition). An optimization analysis shows that T acts as a learned preconditioner within the subspace, yielding spectral-norm step-size bounds and operator-norm variance contraction that stabilize training. Empirically, MTLoRA delivers consistent gains while preserving PEFT efficiency: on GLUE (General Language Understanding Evaluation) with DeBERTaV3-base, MTLoRA improves the average over LoRA by (+2.0) points (86.9 → 88.9) and matches AdaLoRA (88.9) without any pruning schedule; on natural language generation with GPT-2 Medium, it raises BLEU on DART by (+0.95) and on WebNLG by (+0.56); and in multimodal instruction tuning with LLaVA-1.5-7B, DTSM attains the best average (69.91) with  ∼ 4.7% trainable parameters, outperforming full fine-tuning and strong PEFT baselines. These results indicate that learning geometry inside the low-rank subspace improves both effectiveness and stability, making MTLoRA a practical, plug-compatible alternative to LoRA for large-model fine-tuning.

In federated learning (FL), although the original intention of available but not visible data is to allay data privacy concerns, it potentially brings new security threats, particularly poisoning attacks that target such not visible local data. Intuitively, such data poisoning attacks have great potential in stealthily degrading global FL outcomes, and are expected to be even stealthier if being enhanced by generative models like generative adversarial networks (GANs). However, existing defense methods have not been thoroughly challenged in this regard and generally fail to be aware of a local generation of seemingly legitimate poisoned data. With a growing concern on potentially stealthier attacks, in this paper, a cost-effective defense mechanism named Model Consistency-Based Defense (MCD) is proposed, which offers a comprehensive examination of available local models across multiple feature dimensions, providing an indirect yet effective means of identifying hidden data poisoning attackers. To push the limit of MCD against stealthier attacks, we propose a new GAN-based data poisoning attack model named VagueGAN and an unsupervised variant of it, which can be flexibly deployed to generate seemingly legitimate but noisy poisoned data. The consistency of GAN outputs revealed by VagueGAN helps strengthen MCD to work against stealthier GAN-based attacks as well as other mainstream ones. Extensive experiments on multiple open datasets (MNIST, Fashion-MNIST, CIFAR-10, CIFAR-100, and Mini-Imagenet) indicate that our attack method better balances the trade-off between attack effectiveness and stealthiness with low complexity. More importantly, our defense mechanism is shown to be more competent in identifying a variety of poisoned data, particularly stealthier GAN-poisoned ones.

Federated learning (FL) suffers from performance degradation due to the inevitable presence of noisy annotations in distributed scenarios. Existing approaches have advanced in distinguishing noisy samples from the dataset for label correction by leveraging loss values. However, noisy samples recognition relying on scalar loss lacks reliability for FL under heterogeneous scenarios. In this paper, we rethink this paradigm from a representation perspective and propose FedRG( Fed erated under R epresentation G emometry), which follows ''the principle of ``representation geometry priority'' to recognize noisy labels. Firstly, FedRG creates label-agnostic spherical representations by using self-supervision. It then iteratively fits a spherical von Mises-Fisher (vMF) mixture model to this geometry using previously identified clean samples to capture semantic clusters. This geometric evidence is integrated with a semantic-label soft mapping mechanism to derive a distribution divergence between the label-free and annotated label-conditioned feature space, which robustly identifies noisy samples and updates the vMF mixture model with the newly separated clean dataset. Lastly, we employ an additional personalized noise absorption matrix on noisy labels to achieve robust optimization. Extensive experimental results demonstrate that FedRG significantly outperforms state-of-the-art methods for FL with data heterogeneity under diverse noisy client scenarios.

Federated Edge Learning (FEL) has emerged as a promising approach for enabling edge devices to collaboratively train machine learning models while preserving data privacy. Despite its advantages, practical FEL deployment faces significant challenges related to device constraints and device-server interactions, necessitating heterogeneous, user-adaptive model training with limited and uncertain communication. While knowledge cache-driven federated learning offers a promising FEL solution for demanding edge environments, its logits-based interaction design provides poor richness of exchanged information for on-device model optimization. To tackle this issue, we introduce DistilCacheFL, a novel personalized FEL architecture that enhances the exchange of optimization insights while delivering state-of-the-art performance with efficient communication. DistilCacheFL incorporates the benefits of both dataset distillation and knowledge cache-driven federated learning by storing and organizing distilled data as knowledge in the server-side knowledge cache, allowing devices to periodically download and utilize personalized knowledge for local model optimization. Moreover, a device-centric cache sampling strategy is introduced to tailor transferred knowledge for individual devices within controlled communication bandwidth. Extensive experiments on five datasets covering image recognition, audio understanding, and mobile sensor data mining tasks demonstrate that (1) DistilCacheFL significantly outperforms state-of-the-art methods regardless of model structures, data distributions, and modalities. (2) DistilCacheFL can train splendid personalized on-device models with at least 28.6 improvement in communication efficiency.

Parameter efficiency and adaptability are key challenges in fine-tuning large language models (LLMs). Existing parameter-efficient fine-tuning (PEFT) methods such as low-rank adaptation (LoRA) reduce training cost but rely on fixed low-rank adapters, limiting flexibility and underutilizing low-rank structure. We propose triangular adaptive LoRA (TriAdapt-LoRA), a PEFT method that couples a triangular split-based low-rank parameterization with a lightweight importance-driven rank-growth scheme. TriAdapt-LoRA comprises three components: 1) a triangular split-based low-rank structure that enables more flexible modeling within the low-rank subspace; 2) an efficient importance estimator based on changes in the Frobenius norm of transformation matrices, serving as a gradient-informed proxy for module contribution without storing per-step gradients; and 3) a dynamic-rank-growth mechanism that allocates a fixed global rank budget toward more important modules. The design is motivated by the need to allocate limited adaptation capacity to the most useful components during optimization, and is instantiated as a practical, fully gradient-based algorithm. Extensive experiments on natural language understanding, question answering, multimodal reasoning, and multilingual benchmarks show that TriAdapt-LoRA is competitive with strong PEFT baselines and often outperforms AdaLoRA, IncreLoRA, and related methods under matched or reduced adaptation budgets, while reducing the cost of importance estimation, making it a scalable and resource-efficient solution for fine-tuning LLMs.

Content-based image retrieval plays a critical role in diverse applications such as search engines and E-commerce. Nevertheless, a fundamental concern related to label noise within data arises. Label noise is often introduced by imperfect annotation approaches, which mislead retrieval models to memorize spurious patterns and thereby reduce the generalization performance of retrieval models. Firstly, we provide both quantitative and visual evidence that label noise induces dimensional collapse in the learned representation space, manifested as abnormal correlation response. To address this, we propose a robust image retrieval framework TriTAN to mitigate the negative impact caused by label noise while preserving the efficiency. In detail, TriTAN follows a two-stage training scheme. The first stage warms up the model with a noise-robust supervised objective. The second stage couples prototype-guided similarity learning with prototype-based relabeling and explicit representation decorrelation regularization. Experimental results on six datasets demonstrate the robust retrieval robustness of TriTAN compared to state-of-the-art methods under both synthetic and real-world label noise scenarios. We also verify the robust scalability of the proposed framework on diverse vision backbones. Codes will be opensourced to facilitate future research.

The rapid spread of fake news threatens social stability and public trust, highlighting the urgent need for its effective detection. Although large language models (LLMs) show potential in fake news detection, they are limited by knowledge cutoff and easily generate factual hallucinations when handling time-sensitive news. Furthermore, the thinking of a single LLM easily falls into early stance locking and confirmation bias, making it hard to handle both content reasoning and fact checking simultaneously. To address these challenges, we propose ZoFia, a two-stage zero-shot fake news detection framework. In the first retrieval stage, we propose novel Hierarchical Salience and Salience-Calibrated Minimum Marginal Relevance (SC-MMR) algorithm to extract core entities accurately, which drive dual-source retrieval to overcome knowledge and evidence gaps. In the subsequent stage, a multi-agent system conducts multi-perspective reasoning and verification in parallel and achieves an explainable and robust result via adversarial debate. Comprehensive experiments on two public datasets show that ZoFia outperforms existing zero-shot baselines and even most few-shot methods. Our code has been open-sourced to facilitate the research community at https://github.com/SakiRinn/ZoFia.

The rise of End-Edge-Cloud Collaboration (EECC) offers a promising paradigm for Artificial Intelligence (AI) model training across end devices, edge servers, and cloud data centers, providing enhanced reliability and reduced latency. Hierarchical Federated Learning (HFL) can benefit from this paradigm by enabling multi-tier model aggregation across distributed computing nodes. However, the potential of HFL is significantly constrained by the inherent heterogeneity and dynamic characteristics of EECC environments. Specifically, the uniform model structure bounded by the least powerful end device across all computing nodes imposes a performance bottleneck. Meanwhile, coupled heterogeneity in data distributions and resource capabilities across tiers disrupts hierarchical knowledge transfer, leading to biased updates and degraded performance. Furthermore, the mobility and fluctuating connectivity of computing nodes in EECC environments introduce complexities in dynamic node migration, further compromising the robustness of the training process. To address multiple challenges within a unified framework, we propose End-Edge-Cloud Federated Learning with Self-Rectified Knowledge Agglomeration (FedEEC), which is a novel EECC-empowered FL framework that allows the trained models from end, edge, to cloud to grow larger in size and stronger in generalization ability. FedEEC introduces two key innovations: (1) Bridge Sample Based Online Distillation Protocol (BSBODP), which enables knowledge transfer between neighboring nodes through generated bridge samples, and (2) Self-Knowledge Rectification (SKR), which refines the transferred knowledge to prevent suboptimal cloud model optimization. The proposed framework effectively handles both cross-tier resource heterogeneity and effective knowledge transfer between neighboring nodes, while satisfying the migration-resilient requirements of EECC.

Federated learning (FL) is vulnerable to data poisoning attacks due to its distributed nature. Although recent GAN-based data poisoning methods have indicated the potential of using generative AI to generate seemingly legitimate poisoned data, the inherent consistency of GAN outputs can still reveal a sign of data poisoning. In this paper, we propose a diffusion-based data poisoning framework against FL systems, which leverages a Poisoning-Oriented Conditional Diffusion Model (PCDM) to enable fine-grained control over the local generation of poisoned data while ensuring both attack effectiveness and stealthiness. Our PCDM incorporates an adjustable poisoning vector within the global context to precisely control the generation of poisoned data, with theoretical guarantees on attack performance. Furthermore, it employs a novel jumping diffusion strategy for lightweight and efficient poisoned data generation. We conduct the most systematic and broad experimental evaluation for FL poisoning attacks against various defenses, including advanced Byzantine robust aggregation mechanisms, on four open datasets: MNIST, Fashion-MNIST, CIFAR-10, CIFAR-100, and a real-world wireless-specific dataset VRAI. Our results demonstrate that PCDM is less likely to exhibit statistical anomalies compared with the state-of-the-art methods while more effectively degrading global FL performance, which poses a significant risk to data security in FL.

The alignment of large language models (LLMs) with human values is critical for their safe and effective deployment across diverse user populations. However, existing benchmarks often neglect cultural and demographic diversity, leading to limited understanding of how value alignment generalizes globally. In this work, we introduce DiverValue-Bench, a benchmark that systematically evaluates LLMs’ alignment with multi-dimensional human value preferences across 74 countries/regions. DiverValue-Bench contains 23,763 high-quality instances annotated with fine-grained value labels, personalized questions, and rich demographic metadata, providing broad demographic and geographic coverage for population-aware value-alignment evaluation. Using DiverValue-Bench, we conduct an in-depth analysis of several representative LLMs, revealing substantial disparities in alignment performance across geographic and demographic lines. We further demonstrate that lightweight fine-tuning methods, such as Low-Rank Adaptation (LoRA) and Direct Preference Optimization (DPO), can significantly enhance value alignment in both in-domain and out-of-domain settings. Our findings underscore the necessity for population-aware alignment evaluation and provide actionable insights for building culturally adaptive and value-sensitive LLMs. DiverValue-Bench serves as a practical foundation for future research on global alignment, personalized value modeling, and equitable AI development.

Accurate short-horizon vehicle trajectory prediction is critical for autonomous driving and vehicle-to-everything (V2X) communication. Most existing deep learning-based prediction methods rely on fixed-length historical trajectories and multi-agent context, which may be unavailable when a vehicle is newly observed. This paper studies single-frame short-horizon trajectory prediction at signalized intersections, where only one enriched observation frame is used for future motion-state forecasting. We propose a Neural Ordinary Differential Equation (Neural ODE)-based framework that formulates single-frame prediction as continuous-time motion-state forecasting. The proposed model encodes instantaneous kinematic variables, road-context information, and training-set-derived spatial priors into a latent representation, evolves the latent state in continuous time, and decodes future state variables. The predicted motion states are converted into future vehicle coordinates through kinematic integration. Experiments on the CitySIM-Intersection A dataset, evaluated by ADE and DE, show that the proposed method achieves competitive short-horizon prediction accuracy under the single-frame setting. The experiments include baseline comparisons with classical motion models and single-frame neural baselines, input ablation, coordinate and spatial-cell analyses, runtime evaluation, and supplementary studies on multi-frame sequence baselines, longer-horizon rollout, robustness, maneuver-specific performance, and statistical significance. The results clarify the applicability and limitations of single-frame short-horizon prediction on both straight and curved driving subsets.

Remote sensing (RS) applications increasingly operate over distributed infrastructures that integrate space-air-ground-sea resources with edge intelligence, yet remains challenging to centralize due to geographic dispersion, cross-institution barriers and privacy regulations. Federated learning (FL), a promising privacy-preserving distributed learning paradigm, has garnered wide attention. However, the practical application of FL for RS encounters the issue of label noise stemming from inevitable annotation errors. In this work, we pioneer an early investigation of label noise in distributed RS tasks. We introduce the Federated Representation Optimal Matching (FedROM) framework, which guides robust representation alignment in the presence of noisy labels without requiring auxiliary data or transmitting extra sensitive information. Specifically, FedROM focuses on the robust local updating process, where clients first identify underlying noisy samples from the perspectives of both per-sample loss value and latent representation space. Subsequently, inspired by the optimal transport technique, we adaptively align the latent representations of identified noisy samples with their corresponding closest class centroids with the least representation matching distance, where class centroids are averaged by the latent representations of other relatively clean samples. This reduces the misleading effects caused by noisy samples and guides the model to capture more robust semantic features in the latent representation space. Theoretical analysis proves the robustness and convergence of FedROM. Extensive experiments on two real-world distributed RS datasets covering multi-source domains and varying label noise rates demonstrate the robustness of FedROM against eighteen baseline methods. Meanwhile, FedROM also surpasses its counterparts in conditions of no label noise, narrowing the gap with the centralized training.

The rapid advancement of Artificial Intelligence (AI) technology is profoundly transforming human society and concurrently presenting a series of ethical, legal, and social issues. The effective governance of AI has become a crucial global concern. Since 2022, the extensive deployment of generative AI, particularly large language models, marked a new phase in AI governance. Continuous efforts are being made by the international community in actively addressing the novel challenges posed by these AI developments. As consensus on international governance continues to be established and put into action, the practical importance of conducting a global assessment of the state of AI governance is progressively coming to light. In this context, we initiated the development of the AI Governance InternationaL Evaluation Index (AGILE Index). Adhering to the design principle, "the level of governance should match the level of development," the inaugural evaluation of the AGILE Index commences with an exploration of four foundational pillars: the development level of AI, the AI governance environment, the AI governance instruments, and the AI governance effectiveness. It covers 39 indicators across 18 dimensions to comprehensively assess the AI governance level of 14 representative countries globally. The index is utilized to delve into the status of AI governance to date in 14 countries for the first batch of evaluation. The aim is to depict the current state of AI governance in these countries through data scoring, assist them in identifying their governance stage and uncovering governance issues, and ultimately offer insights for the enhancement of their AI governance systems.

For distributed network traffic prediction with data localization and privacy protection, Federated Learning (FL) enables collaborative training without raw data exchange across Base Stations (BSs). Nevertheless, traffic across BSs exhibit inherently heterogeneous trend burst and smooth fluctuation properties, but existing FL methods model single-scale series from only one view, which cannot simultaneously capture diverse trend and fluctuation properties, especially distinct burst distributions. In this paper, we propose Personalized Federated Forecasting with Multi-property Self-fusion (P2FMS), which can represent multi-scale traffic properties from different views. With precise multi-property representations, a fusion-level prediction decision is learned for each client in a personalized manner to promptly sense traffic bursts and improve forecasting performance in non-IID settings. Specifically, P2FMS decomposes the traffic series into distinct time scales, based on which, we effectively extract closeness, period, and trend properties from different views. The closeness and period are embedded through global-view representations with spatial correlations, while non-stationary trends are individually fitted from the client-side view. Furthermore, a personalized combiner is designed to accurately quantify the proportion of general fluctuation raws (i.e., closeness and period) and specific trend property in predictions, which enables multi-property self-fusion for each client to accommodate heterogeneous traffic patterns and enhance prediction accuracy. Besides, an alternant training mechanism is introduced to optimize property representation and fusion modules with the convergence guarantee. Extensive experiments on real-world datasets show that P2FMS outperforms status quo methods in both prediction performance and convergence time.

Benefitting from UAVs' characteristics of flexible deployment and controllable movement in 3D space, odor source localization with multiple UAVs has been a hot research area in recent years. Considering the limited resources and insufficient battery capacities of UAVs, it is necessary to fast locate the odor source with low-complexity computation and minimal interaction under complicated environmental states. To this end, we propose a multi-UAV collaboration based odor source localization (MUC-OSL) method, where source estimation and UAV navigation are iteratively performed, aiming to accelerate the searching process and reduce the resource consumption of UAVs. Specifically, in the source estimation phase, we present a collaborative particle filter algorithm on the basis of UAVs' cognitive difference and collaborative information to improve source estimation accuracy. In the following navigation phase, an adaptive path planning algorithm is designed based on partially observable Markov decision process to distributedly determine the subsequent flying direction and moving steps of each UAV. The results of experiments conducted on two simulation platforms demonstrate that MUC-OSL outperforms existing efforts in terms of mean search time and success rate, and effectively reduces the resource consumption of UAVs.

Federated learning (FL) enables collaborative model training among participants while guaranteeing the privacy of raw data. Mainstream FL methodologies overlook the dynamic nature of real-world data, particularly its tendency to grow in volume and diversify in classes over time. This oversight results in FL methods suffering from catastrophic forgetting, where the trained models inadvertently discard previously learned information upon assimilating new data. In response to this challenge, we propose a novel federated class-incremental learning (FCIL) method, named Federated Class-incremental Learning with New-Class Augmented Self-Distillation (FedCLASS). The core of FedCLASS is to enrich the class scores of historical models with new class scores predicted by current models and utilize the combined knowledge for self-distillation, enabling a more sufficient and precise knowledge transfer from historical models to current models. Theoretical analyses demonstrate that FedCLASS stands on reliable foundations, considering the scores of old classes predicted by historical models as conditional probabilities in the absence of new classes, and the scores of new classes predicted by current models as the conditional probabilities of class scores derived from historical models. Empirical experiments demonstrate the superiority of FedCLASS over four baseline algorithms in reducing average forgetting rate and boosting global accuracy.

Multiple federated learning (FL) methods are proposed for traffic flow forecasting (TFF) to avoid heavy-transmission and privacy-leaking concerns resulting from the disclosure of raw data in centralized methods. However, these FL methods adopt offline learning which may yield subpar performance, when concept drift occurs, i.e., distributions of historical and future data vary. Online learning can detect concept drift during model training, thus more applicable to TFF. Nevertheless, the existing federated online learning method for TFF fails to efficiently solve the concept drift problem and causes tremendous computing and communication overhead. Therefore, we propose a novel method named Resource-Efficient Federated Online Learning (REFOL) for TFF, which guarantees prediction performance in a communication-lightweight and computation-efficient way. Specifically, we design a data-driven client participation mechanism to detect the occurrence of concept drift and determine clients' participation necessity. Subsequently, we propose an adaptive online optimization strategy, which guarantees prediction performance and meanwhile avoids meaningless model updates. Then, a graph convolution-based model aggregation mechanism is designed, aiming to assess participants' contribution based on spatial correlation without importing extra communication and computing consumption on clients. Finally, we conduct extensive experiments on real-world datasets to demonstrate the superiority of REFOL in terms of prediction improvement and resource economization.