Exemplar replay has become an effective strategy for mitigating catastrophic forgetting in federated continual learning (FCL) by retaining representative samples from past tasks. Existing studies focus on designing sample-importance estimation mechanisms to identify information-rich samples. However, they typically overlook strategies for effectively utilizing the selected exemplars, which limits their performance under continual dynamic heterogeneity across clients and tasks. To address this issue, this paper proposes a federated geometry-aware correction method, termed FEAT, which alleviates imbalance-induced representation collapse that drags rare-class features toward frequent classes across clients. Specifically, it consists of two key modules: 1) the Geometric Structure Alignment module performs structural knowledge distillation by aligning the pairwise angular similarities between feature representations and their corresponding Equiangular Tight Frame prototypes, which are fixed and shared across clients to serve as a class-discriminative reference structure. This encourages geometric consistency across tasks and helps mitigate representation drift; 2) the Energy-based Geometric Correction module removes task-irrelevant directional components from feature embeddings, which reduces prediction bias toward majority classes. This improves sensitivity to minority classes and enhances the model's robustness under class-imbalanced data distributions. Extensive experiments on three benchmark datasets demonstrate that FEAT substantially achieves a 4%–8% improvement in Top-1 accuracy compared to nine state-of-the-art methods.

Recent advances in collaborative knowledge distillation have demonstrated cutting-edge performance for resource-constrained distributed multimedia learning scenarios. However, achieving such competitiveness requires addressing a fundamental mismatch: high-dimensional teacher knowledge complexity versus heterogeneous client learning capacities, which currently prohibits deployment in edge-based visual analytics systems. Drawing inspiration from curriculum learning principles, we introduce Federated Adaptive Progressive Distillation (FAPD), a consensus-driven framework that orchestrates adaptive knowledge transfer. FAPD hierarchically decomposes teacher features via PCA-based structuring, extracting principal components ordered by variance contribution to establish a natural visual knowledge hierarchy. Clients progressively receive knowledge of increasing complexity through dimension-adaptive projection matrices. Meanwhile, the server monitors network-wide learning stability by tracking global accuracy fluctuations across a temporal consensus window, advancing curriculum dimensionality only when collective consensus emerges. Consequently, FAPD provably adapts knowledge transfer pace while achieving superior convergence over fixed-complexity approaches. Extensive experiments on three datasets validate FAPD's effectiveness: it attains 3.64

Model-heterogeneous federated learning (MHFL), supporting FL collaboration across clients with heterogeneous models, has become a more practical FL paradigm. Existing MHFL methods enable knowledge fusion over heterogeneous client models by sharing partial homogeneous parameters or extracted label-wise average representations, suffering from model performance bottlenecks and privacy leakage risks. To bridge this gap, we propose a novel model-heterogeneous Federated learning method with homogeneous Representation Subspace Learning (FedRSL) instead of sharing model parameters or representations. In FedRSL, each client’s local heterogeneous model comprises a feature extractor and a prediction header. (1) We construct a homogeneous representation subspace for each client to learn local representation knowledge, and the server aggregates them to generate the global representation subspace for representation knowledge fusion. (2) To facilitate representation learning capability while maintaining efficient communication and computation, we design a lightweight linear model as the homogeneous low-rank linear representation subspace. For each local data sample, its personalized representation extracted by the feature extractor is processed by the global representation subspace to produce the corresponding generalized representation. (3) To effectively bi-transfer global generalized and local personalized knowledge, we reduce the distance between the local personalized representation and the corresponding generalized representation. Experiments on 3 computer vision and 1 natural language processing benchmark datasets over 6 baselines demonstrate that FedRSL obtains state-of-the-art model accuracy (up to 5.51% accuracy improvement) while consuming low communication and computation overheads.

Aligning large language models (LLMs) with human preferences in federated learning (FL) is challenging due to decentralized, privacy-sensitive, and highly non-IID preference data. Direct Preference Optimization (DPO) offers an efficient alternative to reinforcement learning with human feedback (RLHF), but its direct application in FL suffers from severe performance degradation under non-IID data and limited generalization of implicit rewards. To bridge this gap, we propose FedPDPO (Federated Personalized Direct Preference Optimization), a personalized federated framework for preference alignment of LLMs. It adopts a parameter-efficient fine-tuning architecture where each client maintains a frozen pretrained LLM backbone augmented with a Low-Rank Adaptation (LoRA) adapter, enabling communication-efficient aggregation. To address non-IID heterogeneity, we devise (1) the globally shared LoRA adapter with the personalized client-specific LLM head. Moreover, we introduce (2) a personalized DPO training strategy with a client-specific explicit reward head to complement implicit rewards and further alleviate non-IID heterogeneity, and (3) a bottleneck adapter to balance global and local features. We provide theoretical analysis establishing the probabilistic foundation and soundness. Extensive experiments on multiple preference datasets demonstrate state-of-the-art performance, achieving up to 4.80

The rapid evolution of visual generative AI has introduced a wide range of intellectual property risks, spanning the unauthorized learning, reproduction, extraction, misuse, and redistribution of protected data and model assets. To address these risks, a growing body of technical defenses has been proposed. However, existing surveys typically organize this literature by lifecycle stage or technical mechanism, which can obscure the protective intent of different methods. This survey presents a two-dimensional taxonomy for IP protection in visual generative models. The primary axis is a Control Logic View, which classifies methods into Information Exposure Control, Generative Behavior Constraint, and Attribution Accountability according to the risk variable they regulate. The secondary axis distinguishes Data IP from Model IP as cross-cutting asset dimensions. Under this framework, we systematically review protection methods, align evaluation protocols with protection objectives, and discuss open challenges including proactive model-level safeguards, standardized evaluation, robustness against adaptive attacks, and explainable evidence. This survey aims to offer a principled, systematic, and easy-to-follow overview for both new and experienced researchers in visual generative AI IP protection.

Generating high-quality time-series data is challenging because real-world signals often exhibit multimodal patterns and multiscale dynamics, including oscillations and high-frequency variations. Flow Matching (FM) offers an efficient alternative to diffusion models, but practical implementations typically rely on a single finite-capacity global vector-field estimator. In such heterogeneous temporal distributions, distinct regimes may pass through nearby flow states while requiring incompatible conditional velocities. A monolithic estimator trained with the standard ℓ_2 velocity-matching objective may therefore learn an overly smoothed approximation of the local transport field. This estimator-level smoothing can attenuate branch-specific dynamics, leading to spectral distortion and poor mode coverage. To address this, we propose PrismFlow, a new FM method with Koopman-inspired dynamical experts. Each expert learns residual corrections in a latent space where local nonlinear temporal evolution can be approximated by linear transitions. We further propose a confidence-aware Winner-Take-All (WTA) objective that updates only the expert best aligned with each sample while masking gradients to the others, encouraging mode-specific specialization. During sampling, the selected expert adds a residual dynamical correction to the global transport field, preserving FM stability while recovering fine-grained and high-frequency temporal structures. Across various benchmarks, PrismFlow effectively mitigates the spectral contraction in standard FM and achieves state-of-the-art performance, with a 15.6

Cross-market recommendation (CMR) aims to enhance recommendation performance across multiple markets. Due to its inherent characteristics, i.e., data isolation, non-overlapping users, and market heterogeneity, CMR introduces unique challenges and fundamentally differs from cross-domain recommendation (CDR). Existing CMR approaches largely inherit CDR by adopting the one-to-one transfer paradigm, where a model is pretrained on a source market and then fine-tuned on a target market. However, such a paradigm suffers from CH1. source degradation, where the source market sacrifices its own performance for the target markets, and CH2. negative transfer, where market heterogeneity leads to suboptimal performance in target markets. To address these challenges, we propose FeCoSR, a novel federated collaboration framework for cross-market sequential recommendation. Specifically, to tackle CH1, we introduce a many-to-many collaboration paradigm that enables all markets to jointly participate in and benefit from training. It consists of a federated pretraining stage for capturing shared behavior-level patterns, followed by local fine-tuning for market-specific item-level preferences. For CH2, we theoretically and empirically show that vanilla Cross-Entropy (CE) exacerbates market heterogeneity, undermining federated optimization. To address this, we propose a Semantic Soft Cross-Entropy (S^2CE) that leverages shared semantic information to facilitate collaborative behavioral learning across markets. Then, we design a market-specific adaptation module during fine-tuning to capture local item preferences. Extensive experiments on the real-world datasets demonstrate the advantages of FeCoSR over other methods.

In this paper, we aim to address the challenge of hybrid mobile edge-quantum computing (MEQC) for sustainable task offloading scheduling in mobile networks. We develop cost-effective designs for both task offloading mode selection and resource allocation, subject to the individual link latency constraint guarantees for mobile devices, while satisfying the required success ratio for their computation tasks. Specifically, this is a time-coupled offloading scheduling optimization problem in need of a computationally affordable and effective solution. To this end, we propose a deep reinforcement learning (DRL)-based Lyapunov approach. More precisely, we reformulate the original time-coupled challenge into a mixed-integer optimization problem by introducing a penalty part in terms of virtual queues constructed by time-coupled constraints to the objective function. Subsequently, a Deep Q-Network (DQN) is adopted for task offloading mode selection. In addition, we design the Deep Deterministic Policy Gradient (DDPG)-based algorithm for partial-task offloading decision-making. Finally, tested in a realistic network setting, extensive experiment results demonstrate that our proposed approach is significantly more cost-effective and sustainable compared to existing methods.

Federated Class-Incremental Learning (FCIL) aims to continually expand a model’s recognition capacity in a distributed environment, enabling it to learn new classes while retaining knowledge of previously seen ones. Exemplar replay has emerged as a promising strategy owing to its simplicity and effectiveness. Existing methods either select exemplars based on local dynamics or construct global feature spaces to identify representative samples. However, they face inherent challenges in striking a balance between effectiveness and privacy. To address this issue, this paper proposes a Cross-view Lewis weIght Fusion method for exemplar replay in FCIL, termed CLIF, which fuses multi-view importance scores to guide representative sample selection under federated settings. Specifically, CLIF consists of two main modules: 1) the cross-view Lewis weight fusion module computes and integrates Lewis weights from multiple feature perspectives to achieve consistent importance estimation, ensuring that the selected samples better reflect the global data distribution and thus enhancing the representativeness of the replay subset. Building on this, 2) the frequency-based weighted training module adjusts the loss contribution of each sample according to its selection frequency across views, which emphasizes the contribution of critical samples. Moreover, we provide a theoretical analysis to guarantee the soundness and effectiveness of CLIF. Extensive experiments on three datasets demonstrate that our method consistently improves baselines by 1%–6%, supporting the above claims.

Federated learning (FL) is an emerging machine learning paradigm in which a central server coordinates multiple participants (FL clients) collaboratively to train on decentralized data. In practice, FL often faces data, system, and model heterogeneity, which inspires the field of Model-Heterogeneous Personalized Federated Learning (MHPFL). However, existing MHPFL methods rely on extra public data or ignore the relationship between local private heterogeneous models and shared homogeneous models across clients. This leads to unsatisfactory model performance, computational overheads, and communication costs. To bridge this gap, we propose a novel and efficient model-heterogeneous personalized Federated learning framework (pFedLoRA) based on sharing homogeneous Low-Rank Adapter (LoRA) which is popular for fine-tuning pre-trained models. Specifically, we devise a lightweight homogeneous adapter, rather than apply the typical LoRA, to facilitate each client's heterogeneous local model training with our proposed iterative training for global-local bidirectional knowledge exchange. The homogeneous small local adapters are aggregated on the FL server to generate a global adapter. We theoretically prove its $\mathcal {O}(1/T)$ non-convex convergence rate. Experiments on 5 datasets demonstrate pFedLoRA outperforms 9 state-of-the-art baselines in model accuracy with $11.81 \times$ computation and $7.41\times$ communication cost saving.

Small object detection in complex scenes exposes a fundamental tension in neural network design: backbone attention distributes computation uniformly regardless of content, pyramid necks inflate activation magnitudes during upsampling without norm compensation, and bottleneck convolutions progressively smooth high-frequency edge components through accumulated spatial filtering. In response, we develop DFIR-DETR by tracing each proposed module back to a specific, measurable deficiency in the RT-DETR baseline: uniform attention that ignores spatial complexity, norm drift that destabilises upsampled features, and spatial convolutions that progressively suppress the high-frequency components small objects depend on. On NEU-DET and VisDrone, DFIR-DETR achieves 92.9% and 51.6% mAP50 with only 11.7M parameters and 47.2 GFLOPs, demonstrating consistent gains across two qualitatively different detection domains.

With growing client diversity, model-heterogeneous personalized federated learning (MHPFL) supports collaboration over structure-heterogeneous client models. However, existing MHPFL methods only achieve client-level personalization but ignore inherent discrepancies within each client's different data samples, leading to limited model performance. To this end, we propose a novel model-heterogeneous personalized Federated learning with Mixture of Experts (pFedMoE) to achieve a fine-grained data-level personalization. As the first work that incorporates MoE in MHPFL, it introduces three innovations: (1) Different clients hold heterogeneous local models, we add a small proxy global homogeneous feature extractor shared by clients for knowledge exchange. (2) To achieve a fine-grained data-level personalization, we construct a personalized local MoE for each client: a local expert (local heterogeneous client model's feature extractor), a global expert (global proxy homogeneous feature extractor), and a local personalized gating network, which dynamically balances the generalization and personalization of the local model at the data sample level. (3) We customize a lightweight linear gating network to capture the generalized and personalized data characteristics of each local data sample. We theoretically prove its O(1/T) convergence rate. Experiments on 3 benchmark image datasets, 1 real-world image dataset and 1 real-world text dataset against 9 baselines demonstrate its state-of-the-art model accuracy with up to 2.79% accuracy improvement while saving up to 43.12% computational overheads and keeping satisfactory communication costs.

Machine learning, viewed through a causal lens, has sparked a strong interest in the pursuit of fairness. Counterfactual fairness, a popular notion within this framework, evaluates the fairness of a model by comparing its predictions for each individual in the real world with those in counterfactual scenarios, where sensitive attributes (e.g., race, gender, age) are modified. Existing methods for learning counterfactual fairness models rely on having access to complete and clean data. However, real-world data are often of low quality, frequently affected by issues such as missing values. These challenges hinder the training of accurate causal models and mislead the development of counterfactual fairness models. In this paper, we propose the Counterfactual Fair prediction over Missing Data (CFairMD) method to bridge this important gap. CFairMD proposes a causal importance-weighted autoencoder to impute missing data and perform counterfactual inference. It integrates missing data within a joint model framework that captures the relationships between covariates and outcomes. Additionally, we design a counterfactual symmetric function to generate fair representations, complemented by a causally invariant penalty. Theoretical analysis shows that CFairMD achieves counterfactual fairness guarantees in a scalable and flexible manner. Extensive experimental results on four benchmarking datasets show that CFairMD outperforms eight state-of-the-art approaches, with over 22% higher counterfactual fairness gain on average.

Large language models (LLMs) are increasingly deployed as part of compound AI systems which coordinate multiple modules (e.g., retrievers, tools, verifiers) over long-horizon workflows. Although recent frameworks that propagate textual feedback globally (e.g., TextGrad make it feasible to optimize such pipelines, we identify two depth-scaling failure modes in long-horizon agentic workflows: 1) exploding textual gradient, where textual feedback grows exponentially with depth, leading to prohibitively long message and amplifies evaluation biases; and 2) vanishing textual gradient, where limited long-context ability causes models overemphasize recent or early feedback, while compression of lengthy feedback causes downstream messages to lose specificity gradually as they propagate many hops upstream. To mitigate these issues, we introduce Textual Equilibrium Propagation (TEP), a local learning principle inspired by Equilibrium Propagation in energy-based models. TEP includes two phases: 1) a free phase where a local LLM critics iteratively refine prompts until reaching equilibrium (no further improvements are suggested); and 2) a nudged phase which applies proximal prompt edits with bounded modification intensity, using task-level objectives that propagate via forward signaling rather than backward feedback chains. This design supports local prompt optimization followed by controlled adaptation toward global goals without the computational burden and signal degradation of global textual backpropagation. Across long-horizon QA benchmarks and multi-agent tool-use dataset, TEP consistently improves accuracy and efficiency over global propagation methods such as TextGrad, with gains that increase at greater depths, while preserving the practicality of black-box LLM components in deep compound AI system.

We rethink Federated Learning (FL) from a nested learning perspective, framing the core challenge as how to collaboratively learn optimization rules, not just static models, to tackle Non-IID client data. To address this, we propose Federated Nested Learning (FedNL), a novel framework that reformulates FL as a three-level nested optimization system. FedNL embeds Titans-based linear attention into FL, enabling clients to perform lightweight, zero-shot test-time adaptation by treating a delta rule as an online gradient step. Experiments on Non-IID MMLU and long-context benchmarks show that FedNL achieves competitive performance in short-context reasoning, enhances the performance of long-context retrieval and streaming Cross-Entropy, and maintains constant inference memory.

The integration of Foundation Models (FMs) with Federated Learning (FL) presents a transformative paradigm in Artificial Intelligence (AI). This integration offers enhanced capabilities, while addressing concerns of privacy, data decentralization and computational efficiency. This paper provides a comprehensive survey of the emerging field of Federated Foundation Models (FedFM), elucidating their synergistic relationship and exploring novel methodologies, challenges, and future directions that the FL research field needs to focus on in order to thrive in the age of FMs. A systematic multi-tiered taxonomy is proposed, categorizing existing FedFM approaches for model training, aggregation, trustworthiness, and incentivization. Key challenges, including how to enable FL to deal with high complexity of computational demands, privacy considerations, contribution evaluation, and communication efficiency, are thoroughly discussed. Moreover, this paper explores the intricate challenges of communication, scalability and security inherent in training/fine-tuning FMs via FL. It highlights the potential of quantum computing to revolutionize the processes of training, inference, optimization and security. This survey also introduces the implementation requirement of FedFM and some practical FedFM applications. It highlights lessons learned with a clear understanding of our findings for FedFM. Finally, this survey not only provides insights into the current state and challenges of FedFM, but also offers a blueprint for future research directions, emphasizing the need for developing trustworthy solutions. It serves as a foundational guide for researchers and practitioners interested in contributing to this interdisciplinary and rapidly advancing field.