Meta-learning methods are an effective approach to tackling the challenges of few-shot learning. However, current meta-learning approaches based on artificial neural networks do not introduce brain-inspired learning mechanisms, thereby hindering these models from achieving few-shot learning efficiencies and flexibilities comparable to human performance. Furthermore, existing brain-inspired spiking neural network approaches for meta-learning fail to meet Dale’s law constraints on excitation–inhibition balance during synaptic plasticity-driven weight regulation. This limitation compromises both the biological plausibility and the learning capability of spiking meta-learning. This work introduces a Dale’s law-constrained neural excitation–inhibition balance mechanism for meta-learning the initialization of spiking neural networks, constructing internal representations broadly applicable to many few-shot tasks. Specifically, to ensure strict compliance with Dale’s law (neurotransmitter type invariance) during synaptic weight updates, we integrate a modified projected gradient descent method with synaptic plasticity, thereby maintaining the invariance of neural excitatory and inhibitory properties. Building upon this, we incorporate the Dale’s law-constrained excitation–inhibition balance mechanism into spiking meta-learning and propose the excitation–inhibition balance-based spiking meta-initialization (EI-SMI) algorithm. Next, to enhance the stability and generalization capability of spiking meta-initialization, we employ a multistep weighted loss strategy to guide the network’s meta-learning process. Extensive experiments on five benchmark datasets (Omniglot, Mini-ImageNet, CUB, CIFAR-FS, and CWRU) demonstrate that the proposed EI-SMI outperforms early artificial neural network-based meta-learning approaches on certain specialized tasks and surpasses Spiking MAML across all tasks. The experimental results indicate that our method serves as a solid empirical foundation for developing meta-learning algorithms that are both biologically plausible and mechanistically credible.

In modern industrial scenarios, federated learning (FL) has been widely adopted due to its advantages in privacy preservation and distributed modeling. However, existing FL approaches rely on a single update paradigm, which significantly hampers communication efficiency. Moreover, the server preserves historical aggregation states to model causal propagation across training rounds, instead of performing memoryless per-round aggregation. For this end, we propose a self-adapting federated continual learning with gradient causal constraint (GCC-SFCL). First, the selection of synchronous or asynchronous communication across FL rounds is reformulated as continual learning task, where an experience accumulation mechanism records historical aggregation states to model long-term temporal dependencies among client updates. Based on the aggregated context, we further design a self-adapting scheduling strategy driven by multidimensional client state information, enabling the server to dynamically select optimal aggregation. Furthermore, a causality-steering gradient constraint is introduced to explicitly regulate gradient updates, improving the interpretability of the causal model and allowing the capture of fault propagation paths. Experimental results of a simulation case and the real-world case demonstrate that GCC-SFCL significantly improves not only communication efficiency but also the accuracy of fault propagation path identification.

Rapid developments of offshore wind industry offer a strong demand opportunity for offshore wind turbine remote diagnosis. As offshore wind turbines are often located in harsh and communication-constrained environments, the collection and transmission of data is severely restricted, which poses a serious challenge to the conventional centralized diagnostic paradigm that relies on data aggregation. To address this challenge, we propose a novel event-triggered federated learning framework for decentralized fault diagnosis of offshore wind turbines. Specifically, federated learning is first employed to learn decentralized local knowledge from geographically distributed offshore wind turbines, so that the communication objects are transformed from massive raw data into learned parameters, thereby relieving the communication burden. Then, we design an event-triggered communication mechanism and incorporate it into federated learning, the core of which is to modify the communication requirement from uploading all trained parameters periodically to communicating only when necessary. The proposed framework is verified by a real-world offshore wind turbine dataset from six large wind farms in China. An ablation study shows that the proposed framework can maintain high diagnostic performance while reducing communication costs. A comprehensive comparison based on three benchmark models demonstrates that the proposed framework can reduce the communication burden by up to 63% while obtaining better diagnostic performance.Note to Practitioners—This study was motivated by the problem of collaborative diagnosis of distributed offshore wind turbines under the constraints of data privacy and communication overhead. The method employs a federated learning-based fault diagnosis framework, which permits to obtain global fault diagnosis knowledge without aggregating raw data scattered in each end device, thus avoids the risk of data leakage. Moreover, a strategy integrating parameter variation and accuracy gain is designed to avoid communication redundancy for collaborative training. The practicability and superiority of our proposed framework is demonstrated using extensive experiments against actual industrial data collected from six offshore wind farms.

Non-rechargeable batteries remain as the main source of energy for small systems, owing to their unique advantages in energy density, safety, reliability and sustainability. Accurate prediction of the remaining useful life of the battery is not only beneficial to maintenance and production safety, but also can be regarded as a starting point for possible secondary life applications. In this study, an interactive attention sequence-to-sequence network is proposed for the remaining useful life prediction of the non-rechargeable batteries. The proposed approach can effectively extract the degenerate information of each variable-length sequence and dynamically weight the sequence features of different dimensions. For illustration, a case of primary battery dataset collected from the power supply system of 139 vibration sensors is utilized. The extensive experiments verify the effectiveness of the proposed approach.

Privacy protection as a major concern of the industrial big data enabling entities makes the massive safety-critical operation data of a wind turbine unable to exert its great value because of the threat of privacy leakage. How to improve the diagnostic accuracy of decentralized machines without data transfer remains an open issue; especially these machines are almost accompanied by skewed class distribution in the real industries. In this study, a class-imbalanced privacy-preserving federated learning framework for the fault diagnosis of a decentralized wind turbine is proposed. Specifically, a biometric authentication technique is first employed to ensure that only legitimate entities can access private data and defend against malicious attacks. Then, the federated learning with two privacy-enhancing techniques enables high potential privacy and security in low-trust systems. Then, a solely gradient-based self-monitor scheme is integrated to acknowledge the global imbalance information for class-imbalanced fault diagnosis. We leverage a real-world industrial wind turbine dataset to verify the effectiveness of the proposed framework. By comparison with five state-of-the-art approaches and two nonparametric tests, the superiority of the proposed framework in imbalanced classification is ascertained. An ablation study indicates that the proposed framework can maintain high diagnostic performance while enhancing privacy protection.

In this paper, instead of designing new features based on intuition, linguistic knowledge and domain, we learn some new and effective features using the deep autoencoder (DAE) paradigm for phrase-based translation model.Using the unsupervised pre-trained deep belief net (DBN) to initialize DAE's parameters and using the input original phrase features as a teacher for semi-supervised fine-tuning, we learn new semi-supervised DAE features, which are more effective and stable than the unsupervised DBN features.Moreover, to learn high dimensional feature representation, we introduce a natural horizontal composition of more DAEs for large hidden layers feature learning.On two Chinese-English tasks, our semi-supervised DAE features obtain statistically significant improvements of 1.34/2.45(IWSLT) and 0.82/1.52(NIST) BLEU points over the unsupervised DBN features and the baseline features, respectively.

Recursive word topology structure is commonly found in natural language sentences, and discovering this structure can help us to not only identify the units that a sentence contains but also how they interact to form a whole. In this paper, we explore a novel recursive neural network (RNN) based word topology model (WordTM) for hierarchical phrase-based (HPB) speech translation, which captures the topological structure of the words on the source side in a syntactically and semantically meaningful order. Experiments show that our WordTM significantly outperforms the state-of-the-art soft syntactic constraints.