The growing interest in intelligent services and privacy protection for mobile devices has given rise to the widespread application of federated learning Multi-access Edge Computing (MEC). Diverse user behaviors call personalized with heterogeneous Machine Learning (ML) models on different devices. Federated Multi-task (FMTL) is proposed train related but ML devices, whereas previous works suffer from excessive communication overhead during training neglect model heterogeneity among MEC....

Image classification is one of the predominant tasks in computer vision.So far, there are many approaches image classification, and most typical methods Convolutional Neural Networks (CNN), BOF-based algorithms, etc.Most these have a good performance, but still some limitations.Capsule Network (CapsNet) advanced algorithm, which realizes operation based on active vector dynamic routing, can overcome limitations original algorithm.This paper attempts to apply CapsNet into as well another two...

Human embryonic kidney 293T cells (HEK293T cells) before and after treatment with silver nanoparticles (AgNPs) were measured using advanced atomic force microscopy (AFM) measurement technique, the biomechanical property of was analyzed a theoretical model. The results showed that factor viscosity untreated HEK293T reduced from 0.65 to 0.40 for exposure 40 μg/mL AgNPs. Comet assay indicated significant DNA damage occurred in treated cells, as tail DNA% moment. Furthermore, gene expression...

Federated learning (FL) aims to learn joint knowledge from a large scale of decentralized devices with labeled data in privacy-preserving manner. However, noisy labels are ubiquitous reality since high-quality require expensive human efforts, which cause severe performance degradation. Although lot methods proposed directly deal labels, these either excessive computation overhead or violate the privacy protection principle FL. To this end, we focus on issue FL purpose alleviating degradation...

Federated noisy label learning (FNLL) is emerging as a promising tool for privacy-preserving multi-source decentralized learning. Existing research, relying on the assumption of class-balanced global data, might be incapable to model complicated noise, especially in medical scenarios. In this paper, we first formulate new and more realistic federated noise problem where data class-imbalanced heterogeneous, then propose two-stage framework named FedNoRo noise-robust Specifically, stage...

The increasing demand for intelligent services and privacy protection of mobile Internet Things (IoT) devices motivates the wide application Federated Edge Learning (FEL), in which collaboratively train on-device Machine (ML) models without sharing their private data. Limited by device hardware, diverse user behaviors network infrastructure, algorithm design FEL faces challenges related to resources, personalization environments. Fortunately, Knowledge Distillation (KD) has been leveraged as...

Edge Intelligence (EI) allows Artificial (AI) applications to run at the edge, where data analysis and decision-making can be performed in real-time close sources. To protect privacy unify silos distributed among end devices EI, Federated Learning (FL) is proposed for collaborative training of shared AI models across multiple without compromising privacy. However, prevailing FL approaches cannot guarantee model generalization adaptation on heterogeneous clients. Recently, Personalized (PFL)...

Federated Learning (FL) facilitates collaborative model training across decentralized clients, and achieves successes in privacy-sensitive applications such as medical analysis health care. However, data collected annotated by different clients can contain varying degrees of label noise, which decreases the overall convergence leads to performance degradation. We propose a novel end-to-end dual optimization framework, DualOptim, firstly divides into clean noisy groups via analyzing...

Recently, federated learning (FL) has achieved wide successes for diverse privacy-sensitive applications without sacrificing the sensitive private information of clients. However, data quality client datasets can not be guaranteed since corresponding annotations different clients often contain complex label noise varying degrees, which inevitably causes performance degradation. Intuitively, degradation is dominated by with higher rates their trained models more misinformation from data, thus...

<p>Edge Intelligence (EI) enables Artificial (AI) applications to run at the edge, where data analysis and decision-making can be performed in real-time close sources.</p> <p>To protect privacy unify silos distributed among end devices EI, Federated Learning (FL) is proposed for collaborative training shared AI models across multiple without compromising security.</p> <p>However, prevailing FL approaches cannot guarantee model generalization adaptation on...

Federated learning (FL) aims to collaboratively train a shared model across multiple clients without transmitting their local data. Data heterogeneity is critical challenge in realistic FL settings, as it causes significant performance deterioration due discrepancies optimization among models. In this work, we focus on label distribution skew, common scenario data heterogeneity, where the categories are imbalanced each client. To address issue, propose FedBalance, which corrects bias models...

The most expressive way humans display emotions is through facial expressions, and the expression recognition has been widely used. Although so many researches are done, it hard to find a practical application in real world. motion of face modeled by HMM as follows, first, according function processing continuous dynamic signal model recognition, for sample's overlap similarity sample space, Code-HMM was made up respectively; then, inducted KNN some discrimination rules analyzing output...

In the federated learning scenario, geographically distributed clients collaboratively train a global model. Data heterogeneity among significantly results in inconsistent model updates, which evidently slow down convergence. To alleviate this issue, many methods employ regularization terms to narrow discrepancy between client-side local models and server-side However, these impose limitations on ability explore superior ignore valuable information historical models. Besides, although...

Federated Learning (FL) emerges as a distributed machine learning paradigm without end-user data transmission, effectively avoiding privacy leakage. Participating devices in FL are usually bandwidth-constrained, and the uplink is much slower than downlink wireless networks, which causes severe communication bottleneck. A prominent direction to alleviate this problem federated dropout, drops fractional weights of local models. However, existing dropout studies focus on random or ordered lack...

The increasing prevalence of unknown-type attacks on the Internet highlights importance developing efficient intrusion detection systems. While machine learning-based techniques can detect unknown types attacks, need for innovative approaches becomes evident, as traditional methods may not be sufficient. In this research, we propose a deep solution called log-cosh variational autoencoder (LVAE) to address challenge. LVAE inherits strong modeling abilities (VAE), enabling it understand...

Federated Learning (FL) enables collaborative model training among participants while guaranteeing the privacy of raw data. Mainstream FL methodologies overlook dynamic nature real-world data, particularly its tendency to grow in volume and diversify classes over time. This oversight results methods suffering from catastrophic forgetting, where trained models inadvertently discard previously learned information upon assimilating new In response this challenge, we propose a novel...

<p>Federated Learning (FL) enables training Artificial Intelligence (AI) models over end devices without compromising their privacy. As computing tasks are increasingly performed by a combination of cloud, edge, and devices, FL can benefit from this End-Edge-Cloud Collaboration (EECC) paradigm to achieve collaborative device-scale expansion with real-time access. Although Hierarchical Federated (HFL) supports multi-tier model aggregation suitable for EECC, prior works assume the same...

A large amount of sensitive information is generated in today’s evolving network environment. Some hackers utilize low-frequency attacks to steal from users. This generates minority attack samples real traffic. As a result, the data distribution traffic asymmetric, with number normal and rare To address imbalance problem, intrusion detection systems mainly rely on machine-learning-based methods detect attacks. Although this approach can attacks, performance not satisfactory. solve...

With development of the computer technology, large-scale calculation problems are often appeared in network, it needs a lot system resources and support hardware, bring troubles engineering optimization, so apply method such as group's global optimization its improved algorithm to obtain reliable results system. In study, proposes kind particle swarm based on parallel annealing clustering algorithm, is new especially suitable for continuous variable problem. field, can be used computational...

Edge Intelligence (EI) enables Artificial (AI) applications to run at the edge, where data analysis and decision-making can be performed in real-time close sources. To protect privacy unify silos distributed among end devices EI, Federated Learning (FL) is proposed for collaborative training shared AI models across multiple without compromising security. However, prevailing FL approaches cannot guarantee model generalization adaptation on heterogeneous clients. Recently, Personalized (PFL)...