Prototype-based personalized federated learning methods have emerged as a promising strategy due to their ability to represent client-specific class characteristics effectively through learned class prototypes. These prototypes capture salient features of client-local data, facilitating personalized model adaptation. However, existing prototype-based aggregation strategies predominantly rely on weighted averaging, implicitly assuming prototype consistency across clients. This assumption neglects the intrinsic heterogeneity and non-independent and identically distributed (non-IID) nature of client data, compelling diverse local prototypes to align toward a singular global prototype and consequently causing significant aggregation bias. Motivated by observations from intra-class feature saliency analysis, we identify that clients inherently emphasize distinct feature regions even for the same class. To leverage this intra-class diversity, we introduce FedIC, a novel prototype clustering and collaborative classifier optimization approach. Specifically, FedIC first clusters prototypes based on intra-class similarity to form intra-class prototype subspaces, ensuring that aggregation occurs exclusively within each cluster, thus eliminating the bias stemming from forced global unification. To further exploit the benefits of intra-cluster collaboration, we quantify the combined predictive gains of classifiers from clients within the same cluster as a function of classifier combination weights. This targeted aggregation and collaborative optimization strategy effectively circumvents the bias introduced by global alignment. Extensive experiments under various non-IID settings show that FedIC significantly outperforms existing Prototype-based and Clustered PFL Methods.

The limited availability of high-quality training data poses a persistent challenge for synthetic aperture radar (SAR) target classification. Existing data augmentation methods mainly adopt a simplistic application of GAN-based style transfer techniques to directly synthesize pseudo-SAR images from optical images. However, our in-depth analysis of this cross-modal conversion reveals that such straightforward strategies primarily focus on transferring high-level semantic information (e.g., target shapes), thus failing to adequately capture the essential low-level features unique to SAR imagery (e.g., scattering textures). To address this inherent trade-off between high-level semantic preservation and low-level feature authenticity, we propose a Hierarchical Feature-Constrained GAN (HiFC-GAN) tailored for optical-to-SAR style transfer. Specifically, HiFC-GAN enhances the representation of low-level SAR features by introducing local texture contrast constraints at shallow layers, while introducing explicit feature mapping constraints at deeper layers to maintain high-level semantic consistency throughout the reconstruction process. Experimental results demonstrate that HiFC-GAN significantly outperforms existing GAN-based techniques in image generation quality, particularly improving the low-level feature authenticity of pseudo-SAR images. Moreover, the generated pseudo-SAR images further improve the performance of downstream target classification tasks, yielding accuracy gains ranging from 3.56% to 5.90% on average with mainstream CNN-based models.

Negative transfer (NF) is a critical challenge in personalized federated learning (pFL). Existing methods primarily focus on adapting local data distribution on the client side, which can only resist NF, rather than avoid NF itself. To tackle NF at its root, we investigate its mechanism through the lens of the global model, and argue that it is caused by update conflicts among clients during server aggregation. In light of this, we propose a conflict-free client update aggregation strategy (ConFREE), which enables us to avoid NF in pFL. Specifically, ConFREE guides the global update direction by constructing a conflict-free guidance vector through projection and utilizes the optimal local improvements of the worst-performing clients near the guidance vector to regularize server aggregation. This prevents the conflicting components of updates from transferring, achieving balanced updates across different clients. Notably, ConFREE is model-agnostic and can be straightforwardly adopted as a complement to enhance various existing NF-resistance methods implemented on the client side. Extensive experiments demonstrate substantial improvements to existing pFL algorithms by leveraging ConFREE.

Server aggregation conflict is a key challenge in personalized federated learning (PFL). While existing PFL methods have achieved significant progress with shallow base models (e.g., four-layer CNNs), they often overlook the negative impacts of deeper base models on personalization mechanisms. In this paper, we identify the phenomenon of deep model degradation in PFL, where as base model depth increases, the model becomes more sensitive to local client data distributions, thereby exacerbating server aggregation conflicts and ultimately reducing overall model performance. Moreover, we show that these conflicts manifest in insufficient global average updates and mutual constraints between clients. Motivated by our analysis, we proposed a two-stage conflict-aware layer-wise mitigation algorithm (FedCALM), which first constructs a conflict-free global update to alleviate negative conflicts, and then maximizes the benefits of all clients through a conflict-aware strategy. Notably, our method naturally leads to a selective mechanism that balances the tradeoff between clients involved in aggregation and the tolerance for conflicts. Consequently, it can boost the positive contribution to the clients even with the greatest conflicts with the global update. Extensive experiments across multiple datasets and deeper base models demonstrate that FedCALM outperforms four state-of-the-art (SOTA) methods by up to 9.88% and seamlessly integrates into existing PFL methods with performance improvements of up to 9.01%.

Many existing personalized federated learning (PFL) methods utilize clustering-based aggregation to group clients with similar data characteristics, improving model performance by promoting collaboration among clients with shared features. While this method effectively mitigates some challenges posed by data heterogeneity, it predominantly relies on static data features, making it challenging to capture the dynamic changes in client models during iterative training. This limitation impedes accurate clustering based on evolving model updates. To address this issue, we propose a Gradient-Driven Adaptive Clustering method in PFL (GradPFL), which more effectively captures the personalized deviations in locally updated models. Our approach also introduces an adaptive historical gradient mechanism that refines the clustering process by incorporating both current and past update characteristics. This enables more accurate model aggregation that adapts to ongoing changes in client models during training. Experimental results demonstrate that GradPFL outperforms existing clustering-based PFL methods, especially in more complex non-IID environments.

Existing SAR image classification methods based on Contrastive Learning often rely on sample generation strategies designed for optical images, failing to capture the distinct semantic and physical characteristics of SAR data. To address this, we propose Physics-Driven Contrastive Mutual Learning for SAR Classification (PCM-SAR), which incorporates domain-specific physical insights to improve sample generation and feature extraction. PCM-SAR utilizes the gray-level co-occurrence matrix (GLCM) to simulate realistic noise patterns and applies semantic detection for unsupervised local sampling, ensuring generated samples accurately reflect SAR imaging properties. Additionally, a multi-level feature fusion mechanism based on mutual learning enables collaborative refinement of feature representations. Notably, PCM-SAR significantly enhances smaller models by refining SAR feature representations, compensating for their limited capacity. Experimental results show that PCM-SAR consistently outperforms SOTA methods across diverse datasets and SAR classification tasks.

The integration of mobile edge computing (MEC) and heterogeneous networks enables network operators to provide task offloading services to a large number of user devices (UDs) for low-latency task processing by equipping macro base stations and densely deployed small base stations with edge servers. Federated deep reinforcement learning allows each UD to collaboratively learn useful knowledge from the interaction with the environment in a privacy-preserving and high-efficiency way and thus has been applied to solve the task offloading problem in recent studies. However, very few of these studies have considered the energy and time costs incurred by the federated learning process. In this article, the goal is to minimize the total UDs’ energy consumption while guaranteeing deadline constraints considering both the task offloading process and the federated learning process in MEC-enabled heterogeneous networks. Toward this end, we propose a federated deep Q-network (DQN) method where each UD optimizes the offloading decision for the offloading process and the participation decision and training volume for the learning process based on its local DQN model. The simulation results demonstrate the proposed method is superior to several existing methods in terms of energy efficiency and Quality of Service (QoS).

User requests in edge computing environments are inherently decentralized and dynamic, posing significant challenges for efficient and adaptive service replica placement. To address this, we formulate the service replica placement problem in an edge-cloud collaborative environment, explicitly incorporating the spatio-temporal distribution of user requests. By capturing spatial and temporal correlations, we predict future request patterns to enable forward-looking replica placement. Given the NP-hard nature of the optimization problem, we design a DRL algorithm that optimizes replica placement decisions based on predictive modeling. To validate our approach, we conduct extensive experiments on real-world datasets across two typical application scenarios―grid-based and graph-based request distributions. Experimental results show our method reduces average response latency by up to 59.6% and boosts service provider profitability by 4.85% compared to reactive and temporal-only baselines. The proposed framework provides a novel and effective solution for proactive service provisioning in edge computing environments.

In recent years, since edge computing has improved the performance of transportation systems, research on edge-computing-enabled transportation systems has received widespread attention. However, most previous studies overlooked that task requests in transportation systems are unevenly distributed in time and space, which easily causes the overloading of edge servers, resulting in high response latency. To this end, we present a novel task offloading scheme based on graph neural network (GNN) and deep reinforcement learning (DRL) in edge-computing-enabled transportation systems (TransEdge). Specifically, we first propose an adaptive node placement algorithm to assign Internet of Things sensors to appropriate edge servers, thereby minimizing transmission latency. Then, an improved DRL scheme based on GNN is designed to capture the spatial features between sensors, aiming to improve the accuracy of task offloading decisions. Finally, we introduce a task forwarding strategy based on the greedy algorithm to achieve collaborative task offloading between different edge servers and overcome the system instability caused by a sudden surge in task requests. We conduct extensive experiments on two real-world traffic data sets. The results show that TransEdge reduces the response latency by at least 3.7% compared to four baselines while achieving a success rate of 99%.

Recently, the Industrial Internet of Things (IIoT) has shown great application value in environmental monitoring. However, it suffers from serious bottlenecks in energy and computing capability. To address them, researchers have made lots of effort. Nevertheless, they neglect either the edge–end collaboration or the impact of task queue backlog, resulting in low system revenue. To this end, we design a queue-aware computation offloading method based on DRL (QDRL). Specifically, we represent the long-term system operation as a multistage stochastic mixed-integer optimization problem (M-SMIP), which is further converted into a deterministic problem using Lyapunov optimization. Given that the resource allocation and computation offloading in this deterministic problem are strongly coupled and difficult to solve, we decompose this problem into two subproblems. Subsequently, a reinforcement learning scheme with actor–critic architecture is designed to solve these subproblems. The Actor module is designed based on a deep learning model and quantization strategy for generating computation offloading actions. The mathematical reasoning and learning-based methods are integrated as the Critic module for achieving resource allocation. Extensive simulation results show that the performance of QDRL surpasses four baselines and approaches the approximate optimal algorithm in terms of average task queue length, normalized real computation rate, and computation time.

Distributed Machine Learning (DML) techniques are widely used to accelerate the training of large-scale machine learning models. However, during training iterations, gradients need to be frequently aggregated across multiple workers, resulting in communication bottleneck. To reduce the communication overhead of DML, several In-Network Aggregation (INA) protocols are proposed to reduce the volume of aggregation traffic by offloading aggregation functions into switches, thus alleviating network bottlenecks. Nevertheless, these protocols couple the congestion control of in-switch aggregator resources and link bandwidth resources, together with the straggler-oblivious manner in aggregator allocation, leading to low aggregation efficiency.

Multi-feature SAR ship classification aims to build models that can process, correlate, and fuse information from both handcrafted and deep features. Although handcrafted features provide rich expert knowledge, current fusion methods inadequately explore the relatively significant role of handcrafted features in conjunction with deep features, the imbalances in feature contributions, and the cooperative ways in which features learn. In this paper, we propose a novel multi-feature collaborative fusion network with deep supervision (MFCFNet) to effectively fuse handcrafted features and deep features for SAR ship classification tasks. Specifically, our framework mainly includes two types of feature extraction branches, a knowledge supervision and collaboration module, and a feature fusion and contribution assignment module. The former module improves the quality of the feature maps learned by each branch through auxiliary feature supervision and introduces a synergy loss to facilitate the interaction of information between deep features and handcrafted features. The latter module utilizes an attention mechanism to adaptively balance the importance among various features and assign the corresponding feature contributions to the total loss function based on the generated feature weights. We conducted extensive experimental and ablation studies on two public datasets, OpenSARShip-1.0 and FUSAR-Ship, and the results show that MFCFNet is effective and outperforms single deep feature and multi-feature models based on previous internal FC layer and terminal FC layer fusion. Furthermore, our proposed MFCFNet exhibits better performance than the current state-of-the-art methods.

Multi-modal transportation recommendation plays an important role in navigation applications. It aims to recommend a travel plan with various transport modes, such as bus, metro, taxi, bicycle, and a hybrid. Analysis of real-world large-scale navigation data shows that the correlation between the data can be represented by a graph containing different types of nodes and edges. As an emerging technology, graph neural networks (GNN) have shown powerful capabilities in representing graph data. However, existing solutions based on GNN only consider converting heterogeneous graph data into homogeneous graph data, ignoring the effects of different types of nodes and edges. In addition, those methods usually face the over-smoothing problem, which reduces the accuracy of recommendation. To this end, we propose a multi-modalTransportation recommendation algorithm withHeterogeneous graphAttentionNetworks (THAN) based on carefully constructed heterogeneous graphs. We first design a novel graph embedding method to represent the correlation between the origin and the destination, as well as the correlation between origin-destination (OD) pairs and users. Next, a heterogeneous graph from large-scale data is built to describe the relationship between users, OD pairs, and transport modes. Then, we design a hierarchical attention mechanism with residual blocks to generate node embedding in terms of homogeneity and heterogeneity. Finally, a fusion neural layer is designed to fuse embeddings from different views and predict the proper transport mode for users. Extensive experimental results on a large-scale real-world dataset demonstrate that the performance of THAN outperforms five baselines.

Shared e-bikes are popular because of their green, eco-friendly and efficient features. Due to the limited battery capacity of the e-bikes, the energy problem has become one of the main factors limiting its further development. The energy problem can be solved by using static charging piles (SCP) to replenish the batteries of shared e-bike. The location of the shared e-bike is time-varying, resulting in the optimal deployment of SCP as a complex location problem. In this paper, we propose an optimal Deployment algorithm for Maximum Coverage combined the Dynamic Changes of nodes (max-DCDC) based on the known number of SCP. This method first quantitatively analyzes the dynamic change process of the shared e-bike to reduce the deployment scope of the SCP. Then, according to the geometric characteristics of the e-bike distribution within the deployment scope to optimizes the deployment location of the SCP. Simulation experiments show that max-DCDC has better performance in terms of deployment stability and e-bike coverage compared with the other algorithms.