Convolutional Neural Networks (CNN) have been widely adopted in multi-access edge computing system due to their powerful feature extraction and high inference accuracy. Harnessing the heterogeneous computational capacities of end de vices and edge servers, end-edge collaborative inference addresses resource constraints of end devices and latency challenges of real time inference. However, existing studies typically assume that all potential inference models are pre-employed on the server, which may not always align with real-world deployment scenarios. Moreover, the economic incentives of edge service providers are often overlooked in current distributed inference research. To address these issues, we propose an end-edge collaborative inference framework for CNN in multi-access edge computing system, named DisCNN. We first model the interaction between latency-sensitive end devices (EDs) and resource-constrained edge service providers as a Stackelberg game and prove the existence of a Subgame Perfect Equilibrium. Next, we derive the optimal response characteristics of the EDs and develop an efficient algorithm to allocate computation and communication resources to EDs while partitioning the CNN accordingly. To compute the optimal offloading set, we introduce a linear-time algorithm with a bounded approximation ratio. Then, to jointly optimize caching, resource allocation, and pricing, we design an efficient approximate algorithm to compute a near-optimal solution. Finally, extensive simulation results validate the effectiveness of the proposed framework.

Centralized storage in web-based services poses privacy risks for diverse user data, whereas personalized federated learning (PFL) arnesses the inherent privacy-preserving benefits of FL through distributed collaborative training, avoiding the centralization of sensitive data. Besides, PFL also customizes models to individual client needs, enabling personalized web services on non-independent and identically distributed data across varied environments. Existing PFL techniques often rely on prior knowledge to identify personalization layers, overlooking variable layer sensitivities to heterogeneous data, thus constraining the model's adaptability to diverse statistical distributions and limiting its service effectiveness. To overcome this issue, we introduce FedLD, a layer-wise decoupling method for PFL that precisely captures layer-specific sensitivities to heterogeneous data, enhancing service-oriented model personalization. Firstly, FedLD employs a hypernetwork to evaluate each layer's contribution to performance across varied data distributions, determining the optimal proportion of personalized channels per layer for each client, thus enabling fine-grained, channel-level parameter adjustments tailored to individual service needs. Secondly, knowledge distillation is introduced to enforce consistency between the representations of personalized and shared weights, promoting collaboration between them. Thirdly, an improved aggregation strategy is proposed to enhance collaboration among clients with similar data distributions. Extensive experimental results show that FedLD achieves more effective and generalized results compared to baseline methods across various types of heterogeneous data settings.

Seeking to address the challenges associated with high-dimensional complex time series representations of recurrent neural networks, such as low generalization ability and long training time, a hybrid neural network based on a deep belief network (DBN) is proposed in this paper to facilitate time series predictions for the Internet of Things. In our approach, we integrate both a DBN and a recurrent neural network with the gated recurrent unit as the activation unit. First, we implement unsupervised pretraining through the DBN and then supervise the curve fitting using the recurrent neural network. Finally, the hybrid neural network is learned and can make predictions. The experimental results show that the hybrid neural network has a stronger historical learning ability than two other widely used recurrent neural networks and can effectively reduce the number of iterations required by the recurrent neural network, thereby reducing the overall learning time.

Due to the problem of attribute redundancy in meteorological data from the Industrial Internet of Things (IIoT) and the slow efficiency of existing attribute reduction algorithms, attribute reduction based on a genetic algorithm for the coevolution of meteorological data was proposed. The evolutionary population was divided into two subpopulations: one subpopulation used elite individuals to assist crossover operations to increase the convergence speed of the algorithm, and the other subpopulation balanced the population diversity in the evolutionary process by introducing a random population; these two subpopulations completed the evolutionary operations together. With the TSDPSO-AR algorithm and ARAGA algorithm, the attribute reduction operation for precipitation in meteorological data was performed. The results showed that the proposed algorithm maintained the diversity of the population during evolution, improved the reduction performance, and simplified the information system.

Because the existing approaches for diagnosing sensor networks lead to low precision and high complexity, a new fault detection mechanism based on support vector regression and neighbor coordination is proposed in this work. According to the redundant information about meteorological elements collected by a multisensor, a fault prediction model is built using a support vector regression algorithm, and it achieves residual sequences. Then, the node status is identified by mutual testing among reliable neighbor nodes. Simulations show that when the sensor fault probability in wireless sensor networks is 40%, the detection accuracy of the proposed algorithm is over 87%, and the false alarm ratio is below 7%. The detection accuracy is increased by up to 13%, in contrast to other algorithms. This algorithm not only reduces the communication to sensor nodes but also has a high detection accuracy and a low false alarm ratio. The proposed algorithm is suitable for fault detection in meteorological sensor networks with low node densities and high failure ratios.