Diabetic retinopathy (DR) is a debilitating ocular complication demanding timely intervention and treatment. The rapid evolution of deep learning (DL) has notably enhanced the efficiency of conventional manual diagnosis. However, the scarcity of existing DR datasets hinders the progress of data-driven DL models, especially for pixel-level lesion annotation datasets, which severely impedes the advancement of DR lesion segmentation tasks required for precise interpretations of DR grading. Furthermore, the escalating concerns surrounding medical data security and privacy induce data collection challenges for traditional centralized learning, exacerbating the issue of data silos. Federated learning (FL) emerges as a privacy-preserving distributed learning paradigm. Nevertheless, the existing literature lacks a comprehensive FL framework for DR diagnosis and fails to exploit multiple diverse DR datasets simultaneously. To address the challenges of data scarcity and privacy, we construct a high-quality pixel-level DR lesion annotation dataset (TJDR) and propose a novel FL-based DR diagnosis framework including both DR grading and multi-lesion segmentation. Moreover, to tackle the scarcity of pixel-level DR lesion datasets, we propose$\bm {\alpha }$-Fed and adaptive-$\bm {\alpha }$-Fed, two efficient cross-dataset FL algorithms. Extensive experiments demonstrate the effectiveness of our proposed framework and the two cross-dataset FL algorithms.

The evolving Industrial Internet of Things (IIoT) is shifting towards decentralized collaborative manufacturing, posing heightened network security issues within interconnected value chains, thus requiring advanced Network Intrusion Detection (NID) systems to identify potential threats. In this context, traditional centralized NID systems are insufficient due to cross-industrial privacy concerns and interconnected secure threats. Federated Learning (FL) has emerged as a promising solution to enable the sharing of security insights without compromising privacy across participants. However, establishing an FL-based NID framework in realistic IIoT scenarios faces several hurdles, including the limited availability of large-scale devices and heterogeneous attack data distributions. The former leads to inconsistent client participation and degraded performance, while the latter hinders model convergence. To address these, we propose a novel Federated Learning-based Industrial Network Intrusion Detection (FedIn-NID) framework, incorporating a multidimensional client selection strategy and a dynamic global aggregation strategy. The selection strategy synergistically considers multidimensional factors including client availability, local dataset distribution, and dataset size. This approach accommodates clients with varying availability and avoids the selection of biased clients with data concentrated in a few categories. During model aggregation, the proposed strategy leverages the concept of exponential moving average to dynamically balance the holistic yet slightly older knowledge in the global model with the partial but relatively newer knowledge in local models, ensuring effective aggregation and convergence of the global NID model. Experiments demonstrate that FedIn-NID outperforms baselines by 10% to 30%, showcasing remarkable robustness with increasing data distribution heterogeneity and device count.

Federated semi-supervised learning (FSSL) has recently emerged as a promising approach for enhancing the performance of federated learning (FL) using ubiquitous unlabeled data. However, this approach encounters challenges when learning a global model using both fully labeled and fully unlabeled clients. Previous works overlook the dissimilarities between labeled and unlabeled clients, predominantly using shared parameters for local training across these two types of clients, thereby inducing intertask interference during local training. Moreover, these works typically adopt a single-perspective aggregation strategy, primarily focusing on data-volume-aware aggregation (i.e., FedAvg), leading to a lack of comprehensive consideration in model aggregation. In this article, we propose a novel FSSL method termed $\text {D}^{{2}}\text {Fed}$ , which addresses these issues by rethinking the roles of labeled clients and unlabeled ones to mitigate intertask interference during local training and by integrating client-type-aware with data-volume-aware to provide a more comprehensive perspective for model aggregation. Specifically, in local training, our proposed $\text {D}^{{2}}\text {Fed}$ distinguishes between the primary and accessory roles of labeled and unlabeled clients, respectively, performing dual-role additive local training (DALT) accordingly. In global aggregation, $\text {D}^{{2}}\text {Fed}$ uses a dual-perspective global aggregation (DGA) strategy, transitioning from data-volume-aware aggregation to client-type-aware aggregation. The proposed method simultaneously improves both local training and global model aggregation for FSSL without compromising privacy. We demonstrate the effectiveness and robustness of the proposed method through extensive experiments and elaborate ablation studies conducted on the CIFAR-10/100, SVHN, FMNIST, and STL-10 datasets. Experimental results show that $\text {D}^{{2}}\text {Fed}$ outperforms state-of-the-arts on five datasets under diverse data settings.

The rise of collaborative manufacturing, driven by the rapid proliferation of Industrial Internet of Things (IIoT) technologies, has markedly enhanced agility and productivity in industrial environments. However, this advancement has also significantly broadened the attack surface and uncovered unique vulnerabilities intrinsic to these interconnected systems. This paper introduces a novel Hierarchical Federated Incremental Learning Network Intrusion Detection (HFIN) approach. To our knowledge, this is the first attempt to address the ever-evolution network intrusion detection (NID) challenges in IIoT landscapes from the continuous attack-defense perspective. Our proposed HFIN capitalizes on decentralized model training across multifarious IIoT devices, ensuring data privacy and empowering continuous learning capabilities. It utilizes distributed data sources for secure experience sharing, collaboratively enhancing the continuous detection performance of the global model. Furthermore, regarding the inherent resource constraints of IIoT devices, we proposed a novel edge-client Weighted Transmission Optimization strategy (WTO). This strategy adeptly balances effective intrusion detection with the operational constraints of IIoT devices. By holistically considering detection capabilities and data volume across different attack types, it prioritizes the transmission of more critical and scarce attack data for training within bandwidth constraints. This maintains the comprehensive detection capability of the global model against various network attacks. To validate the effectiveness of HFIN, we conduct extensive experiments using the NF-UQ-NIDS-v2 and NF-ToN-IoT-v2 datasets. Experimental results demonstrate that our method outperforms baselines by approximately 10% in terms of accuracy and F1-score, highlighting the applicability and effectiveness of HFIN in enhancing security against sophisticated industrial environments and ever-evolving cyber threats.