This article studies the joint optimization problem of computation offloading and resource allocation (JCORA) in mobile-edge computing (MEC). Deep reinforcement learning (DRL) is one ideal techniques for addressing dynamic JCORA problem. However, it still challenging to adapt traditional DRL methods since they usually lead slow unstable convergence model training. To this end, we propose a temporal attentional deterministic policy gradient (TADPG) tackle JCORA. Based on deep (DDPG), TADPG...

This paper proposes an efficient federated distillation learning system (EFDLS) for multi-task time series classification (TSC). EFDLS consists of a central server and multiple mobile users, where different users may run TSC tasks. has two novel components: feature-based student-teacher (FBST) framework distance-based weights matching (DBWM) scheme. For each user, the FBST transfers knowledge from its teacher's hidden layers to student's via distillation, teacher student have identical...

Over the years, a number of semisupervised deep-learning algorithms have been proposed for time-series classification (TSC). In deep learning, from point view representation hierarchy, semantic information extracted lower levels is basis that higher levels. The authors wonder if high-level also helpful capturing low-level information. This paper studies this problem and proposes robust model with self-distillation (SD) simplifies existing learning (SSL) techniques TSC, called SelfMatch....

Recently, contrastive learning (CL) is a promising way of discriminative representations from time series data. In the representation hierarchy, semantic information extracted lower levels basis that captured higher levels. Low-level essential and should be considered in CL process. However, existing algorithms mainly focus on similarity high-level information. Considering low-level may improve performance CL. To this end, we present deep with self-distillation (DCRLS) for domain. DCRLS...

This article proposes a semi-supervised contrastive capsule transformer method with feature-based knowledge distillation (KD) that simplifies the existing semisupervised learning (SSL) techniques for wearable human activity recognition (HAR), called CapMatch. CapMatch gracefully hybridizes supervised and unsupervised to extract rich representations from input data. In learning, leverages pseudolabeling, (CL), KD construct similarity on lower higher level semantic information extracted two...

Multivariate time series classification (MTSC) based on deep learning (DL) has attracted increasingly more research attention. The performance of a DL-based MTSC algorithm is heavily dependent the quality learned representations providing semantic information for downstream tasks, e.g., classification. Hence, model's representation ability critical enhancing its performance. This article proposes densely knowledge-aware network (DKN) MTSC. DKN's feature extractor consists residual multihead...

This paper proposes a dual-network-based feature extractor, perceptive capsule network (PCapN), for multivariate time series classification (MTSC), including local (LFN) and global relation (GRN). The LFN has two heads (i.e., Head_A Head_B), each containing squash CNN blocks one dynamic routing block to extract the features from data mine connections among them. GRN consists of capsule-based transformer capture patterns variable correlate useful information multiple variables. Unfortunately,...

Over the years, many deep learning algorithms have been developed for time series classification (TSC). A model's performance usually depends on quality of semantic information extracted from lower and higher levels within representation hierarchy. Efficiently promoting mutual between is vital to enhance during model learning. To this end, we propose a self-bidirectional decoupled distillation (Self-BiDecKD) method TSC. Unlike most self-distillation that transfer target-class knowledge...

Mobile edge computing (MEC) has been an effective paradigm for supporting computation-intensive applications by offloading resources at network edge. Especially in vehicular networks, the MEC server, is deployed as a small-scale computation server roadside and offloads task to its local server. However, due unique characteristics of including high mobility vehicles, dynamic distribution vehicle densities heterogeneous capacities servers, it still challenging implement efficient mechanism...

Kriging models, also known as Gaussian process are widely used in surrogate-assisted evolutionary algorithms (SAEAs). However, the cubic time complexity of standard models limits their usage high-dimensional optimization. To tackle this problem, we propose an incremental model for computation. The main idea is to update incrementally based on equations previously trained instead building from scratch when new samples arrive, so that updating can be reduced quadratic. proposed learning scheme...

Mobile edge computing (MEC) has been an effective paradigm to support real-time computation-intensive vehicular applications. However, due highly dynamic topology, these existing centralized-based or distributed-based scheduling algorithms requiring high communication overhead, are not suitable for task offloading in networks. Therefore, we investigate a novel service scenario of MEC-based crowdsourcing, where each MEC server is independent agent and responsible making processing traffic...

This paper studies the trajectory control and task offloading (TCTO) problem in an unmanned aerial vehicle (UAV)-assisted mobile edge computing system, where a UAV flies along planned to collect computation tasks from smart devices (SDs). We consider scenario that SDs are not directly connected by base station (BS) has two roles play: MEC server or wireless relay. The makes decisions online, which collected can be executed locally on offloaded BS for remote processing. TCTO involves...

Abstract Spectrum sensing is an efficient technology for addressing the shortage of spectrum resources. Widely used methods usually employ model‐based features as test statistics, such energies and eigenvalues, ignoring temporal correlation aspect. Deep learning based have potential to focus on various aspects, including correlation. However, existing ones are not good at capturing from data traditional convolutional neural network (CNN) long short‐term memory (LSTM) feature extraction....

This paper investigates the problem of energy-efficient trajectory optimization with wireless charging (ETWC) in an unmanned aerial vehicle (UAV)-assisted mobile edge computing system. A UAV is dispatched to collect computation tasks from specific ground smart devices (GSDs) within its coverage while transmitting energy other GSDs. In addition, a high-altitude platform laser beam deployed stratosphere charge UAV, so as maintain flight mission. The ETWC characterized by multi-objective...

This paper presents a modified ant colony optimization (ACO) approach for the network coding resource minimization problem. It is featured with several attractive mechanisms specially devised solving concerned problem: 1) multidimensional pheromone maintenance mechanism put forward to address issue of overlapping; 2) problem-specific heuristic information employed enhance capability search (neighboring area search); 3) tabu-table-based path construction method facilitate feasible...