Accurate Short-Term Load Forecasting (STLF) is a critical task in managing and operating smart grids. Existing STLF methods primarily rely on mathematical modeling or neural networks, often struggle to effectively capture the correlations between influencing factors and load data, and frequently lack interpretability. To address these challenges, this paper proposes an intelligent framework for STLF that combines a pattern extraction and attention mechanism, which leverages the characteristics of electricity consumption data. The proposed framework facilitates the integration of prior knowledge, identifies intrinsic data patterns, and more accurately maps the relationships between influencing factors and load patterns. Finally, we conduct experiments on real-world and publicly available datasets to evaluate the performance of the proposed model. Specifically, the proposed model improves the accuracy of STLF compared to that of existing methods and reduces the mean absolute percentage error by $2\%$ to $5\%$. The model performs superiorly on the real datasets, with root mean squared error and mean absolute percentage error values of $0.810$ MWh and $7.09\%$.

In smart homes, the computational offloading technology of edge cloud computing (ECC) can effectively deal with the large amount of computation generated by smart devices. In this paper, we propose a computational offloading strategy for minimizing delay based on the back-pressure algorithm (BMDCO) to get the offloading decision and the number of tasks that can be offloaded. Specifically, we first construct a system with multiple local smart device task queues and multiple edge processor task queues. Then, we formulate an offloading strategy to minimize the queue length of tasks in each time slot by minimizing the Lyapunov drift optimization problem, so as to realize the stability of queues and improve the offloading performance. In addition, we give a theoretical analysis on the stability of the BMDCO algorithm by deducing the upper bound of all queues in this system. The simulation results show the stability of the proposed algorithm, and demonstrate that the BMDCO algorithm is superior to other alternatives. Compared with other algorithms, this algorithm can effectively reduce the computation delay.

The detection of pests and diseases in crops is currently a hot topic. The complexity of pest and disease object in the field, combined with inconsistent features across different levels, poses challenges for network detection. Additionally, the complex agricultural production environment tends to generate many interfering negative samples, which significantly complicates pest and disease differentiation. To address these two issues, the YOLOv7‐PSAFP network structure was first proposed. Based on YOLOV7, the progressive Spatial Adaptive Feature Pyramid (PSAFP) was introduced. Second, a combination of the Varifocal Loss and Loss Rank Mining loss functions was used for calculating the object loss, which reduces the interference of useless negative examples during training. On the filtered‐plant‐village‐dataset and rice‐corn pest dataset, the mAP results of YOLOv7‐PSAFP were 84.7 and 93.3, which are 2.9 and 2.1 higher than the baseline model (YOLOv7), respectively. The code for this paper is located at https://github.com/DuLJ72/PSAFP .

Federated learning, as a distributed machine learning framework, aims to protect data privacy while addressing the issue of data silos by collaboratively training models across multiple clients. However, a significant challenge to federated learning arises from the non-independent and identically distributed (non-iid) nature of data across different clients. non-iid data can lead to inconsistencies between the minimal loss experienced by individual clients and the global loss observed after the central server aggregates the local models, affecting the model’s convergence speed and generalization capability. To address this challenge, we propose a novel federated learning algorithm based on update bias (FedUB). Unlike traditional federated learning approaches such as FedAvg and FedProx, which independently update model parameters on each client before direct aggregation to form a global model, the FedUB algorithm incorporates an update bias in the loss function of local models—specifically, the difference between each round’s local model updates and the global model updates. This design aims to reduce discrepancies between local and global updates, thus aligning the parameters of locally updated models more closely with those of the globally aggregated model, thereby mitigating the fundamental conflict between local and global optima. Additionally, during the aggregation phase at the server side, we introduce a metric called the bias metric, which assesses the similarity between each client’s local model and the global model. This metric adaptively sets the weight of each client during aggregation after each training round to achieve a better global model. Extensive experiments conducted on multiple datasets have confirmed the effectiveness of the FedUB algorithm. The results indicate that FedUB generally outperforms methods such as FedDC, FedDyn, and Scaffold, especially in scenarios involving partial client participation and non-iid data distributions. It demonstrates superior performance and faster convergence in tasks such as image classification.

The advancement of deep neural networks (DNNs) has prompted many cloud service providers to offer deep learning as a service (DLaaS) to users across various application domains. However, in current DLaaS prediction systems, users’ data are at risk of leakage. Homomorphic encryption allows operations to be performed on ciphertext without decryption, which can be applied to DLaaS to ensure users’ data privacy. However, mainstream homomorphic encryption schemes only support homomorphic addition and multiplication, and do not support the ReLU activation function commonly used in the activation layers of DNNs. Previous work used approximate polynomials to replace the ReLU activation function, but the DNNs they implemented either had low inference accuracy or high inference latency. In order to achieve low inference latency of DNNs on encrypted data while ensuring inference accuracy, we propose a low-degree Hermite deep neural network framework (called LHDNN), which uses a set of low-degree trainable Hermite polynomials (called LotHps) as activation layers of DNNs. Additionally, LHDNN integrates a novel weight initialization and regularization module into the LotHps activation layer, which makes the training process of DNNs more stable and gives a stronger generalization ability. Additionally, to further improve the model accuracy, we propose a variable-weighted difference training (VDT) strategy that uses ReLU-based models to guide the training of LotHps-based models. Extensive experiments on multiple benchmark datasets validate the superiority of LHDNN in terms of inference speed and accuracy on encrypted data.