The sixth-generation network (6G) is expected to achieve a fully connected world, which makes full use of large amount sensitive data. Federated Learning (FL) an emerging distributed computing paradigm. In Vehicular Edge Computing (VEC), FL used protect consumer data privacy. However, using in VEC will lead expensive communication overheads, thereby occupying regular resources. the traditional FL, massive rounds before convergence enormous costs. Furthermore, each round, many clients upload...

The most significant technical challenges of current aerial image object-detection tasks are the extremely low accuracy for detecting small objects that densely distributed within a scene and lack semantic information. Moreover, existing detectors with large parameter scales unsuitable scenarios oriented toward low-end GPUs. To address this challenge, we propose efficient-lightweight You Only Look Once (EL-YOLO), an innovative model overcomes limitations GPU orientation. EL-YOLO surpasses...

With the development of technologies deployed on vehicles, there is a significant increase in amount data, which comes from various applications, such as battery management, VR, autopilot, etc. However, privacy critical obstacle to utilizing information since many vehicle-based applications involve locations, conversations, driving behaviors, Federated Learning (FL) promising technology perfect for filling gap, it keeps users' data their devices, gives rise Vehicular Networks (FVN). As...

Abstract The research and application of bearing fault diagnosis techniques are crucial for enhancing equipment reliability, extending lifespan, reducing maintenance expenses. Nevertheless, most existing methods encounter challenges in discriminating between signals from machines operating under normal faulty conditions, leading to unstable detection results. To tackle this issue, the present study proposes a novel approach based on graph neural networks ensemble learning. Our key...

To solve the problems of a poor manual garbage sorting environment, including heavy tasks and low efficiency, we propose Lightweight Feature Fusion Single Shot Multibox Detector (L-SSD) algorithm to realize intelligent trash classification recognition. Since waste has small volume image resolution is always low, that an enhanced single shot multibox detector (SSD) with lightweight novel feature fusion module. This SSD can significantly improve performance rubbish detection. In this module,...

Summary Cloud computing is the key and frontier field of current domestic international computer technology, workflow task scheduling plays an important part cloud computing, which a policy that maps tasks to appropriate resources execute. Effective essential for obtaining high performance in environment. In this paper, we present algorithm based on resources' fuzzy clustering named FCBWTS. The major objective minimize makespan precedence constrained applications, can be modeled as directed...

To address the problem of inadequate feature extraction and binary code discrete optimization faced by deep hashing methods using a relaxation-quantization strategy, novel attention-guided method with pairwise labels (DAHP) is proposed to enhance global fusion, better learn contextual information image features effectively representation, solve losing in optimizing loss function. First, we introduce new concept called anchor hash generation(AHCG) algorithm, train ResNet position attention...

Remote sensing image object detection holds significant research value in resources and the environment. Nevertheless, complex background information considerable size differences between objects remote images make it challenging. This paper proposes an efficient model (MSA-YOLO) to improve performance. First, we propose a Multi-Scale Strip Convolution Attention Mechanism (MSCAM), which can reduce introduction of noise fuse multi-scale features enhance focus on foreground various sizes....

The recently developed vision transformer (ViT) has achieved promising results on image retrieval compared to convolutional neural networks. However, most of these transformer-based methods use the original ViT model extract global features, ignoring importance local features for retrieval. In this work, we propose a multiscale feature fusion method (MSViT) achieve with features. challenge research work is how learn representation ability model, so as improve performance model. First,...

For the cloud computing, task scheduling problems are of paramount importance. It becomes more challenging when takes into account energy consumption, traditional make span criteria and users QoS as objectives. This paper considers independent tasks in computing a bi-objective minimization problem with consumption criteria. We use Dynamic Voltage Scaling (DVS) to minimize propose two algorithms. These algorithms methods unify double fitness define function select individuals. They adopt...