In this paper, we propose a novel convolutional neural network based on adaptive multi-scale feature aggregation and boundary-aware for lateral ventricle segmentation (MB-Net), which mainly includes three parts, i.e.,...
详细信息
Robust principal component analysis (RPCA) is widely studied in computer vision. Recently an adaptive rank estimate based RPCA has achieved top performance in low-level vision tasks without the prior rank, but both th...
Robust principal component analysis (RPCA) is widely studied in computer vision. Recently an adaptive rank estimate based RPCA has achieved top performance in low-level vision tasks without the prior rank, but both the rank estimate and RPCA optimization algorithm involve singular value decomposition, which requires extremely huge computational resource for large-scale matrices. To address these issues, an efficient RPCA (eRPCA) algorithm is proposed based on block Krylov iteration and CUR decomposition in this paper. Specifically, the Krylov iteration method is employed to approximate the eigenvalue decomposition in the rank estimation, which requires $O(ndrq+n(rq)^{2})$ for an $(n\times d)$ input matrix, in which $q$ is a parameter with a small value, $r$ is the target rank. Based on the estimated rank, CUR decomposition is adopted to replace SVD in updating low-rank matrix component, whose complexity reduces from $O(rnd)$ to $O(r^{2}n)$ per iteration. Experimental results verify the efficiency and effectiveness of the proposed eRPCA over the state-of-the-art methods in various low-level vision applications.
Remaining useful life (RUL) prediction is highly demanded in modern industry as it provides a scheduling basis for predictive maintenance. Recently, intelligent data-driven methods have been developed for RUL predicti...
详细信息
Remaining useful life (RUL) prediction is highly demanded in modern industry as it provides a scheduling basis for predictive maintenance. Recently, intelligent data-driven methods have been developed for RUL prediction due to their powerful performance. However, existing methods mostly ignore physical dynamics behind the RUL prediction problem, which predict a fluctuated RUL trajectory contrary to the physical intuition. To address this issue, we formulize RUL pre-diction as a time-varying trajectory modeling problem by analyzing the difference between sto-chastic degradation process and smooth RUL trajectory, and propose a dynamic governing network (DGN) to identify the RUL trajectory from life-span observation series. Specifically, a discretized ordinary differential equation (ODE) parameterized by neural networks is utilized to describe a governing equation of the RUL trajectory. To constraint the trajectory space, a nonnegative bounded function is inserted into each time step of the forward propagation of the ODE. To identify time-varying coefficients in the DGN, the ODE network is specified as a super -network with time-invariant parameters and a time-varying network architecture, which is dynamically determined by a deep reinforcement learning algorithm. Experimental results on two datasets demonstrate that the proposed DGN can capture underlying dynamics from observation series and can obtain state-of-the-art RUL prediction performance.
As an emerging metric for information freshness, AoI (Age of Information) is widely used in time-sensitive systems. As a time-sensitive system, the IoT (Internet of Things) system allocates link resources to source no...
详细信息
As an emerging metric for information freshness, AoI (Age of Information) is widely used in time-sensitive systems. As a time-sensitive system, the IoT (Internet of Things) system allocates link resources to source nodes for data transmission so that users can obtain fresh data and quality services, which is particularly important. However, in the existing research work, for the IoT system composed of multiple sources and a common base station, the algorithms adopted only consider minimizing AoI and ignore the existence of sample extrusion phenomenon. Sample extrusion is a phenomenon that occurs when the collected sample stored in the buffer of the source node has not been completely transmitted to the base station, and the source node has collected another new sample. Based on a greedy strategy, we propose a link resource allocation algorithm that considers both AoI and sample extrusion to collect the data from various source nodes. Simulation experiments show that the proposed algorithm can achieve better comprehensive performance than the other existing algorithm.
The text-to-SQL task is a realistic and challenging job, which aims to translate natural language questions into corresponding SQL queries. When attempting to generate SQL queries in the text-to-SQL task, prevailing s...
The text-to-SQL task is a realistic and challenging job, which aims to translate natural language questions into corresponding SQL queries. When attempting to generate SQL queries in the text-to-SQL task, prevailing semantic parsing models employ beam search to generate several candidates. Based on our pilot study, we observe that the gold-standard SQL answer may exist in the n-best candidate list produced by the decoder, rather than the first. Hence, we aim to lift the correct answer given several candidates generated by semantic parsing models for text-to-SQL. To this end, we propose a reranking module that reorders the n-best list of candidate SQL queries by pair-wise hinge loss. Meanwhile, a data augmentation module is leveraged to enrich the inadequate training instances for providing candidates with better qualities. These two modules can be easily grafted onto text-to-SQL backbone networks, and extensive experiments on the cross-domain text-to-SQL benchmark Spider demonstrate that our method achieves 73.8% in accuracy on the Spider dataset, surpassing the base model by up to 3.5%.
作者:
Ma, JianfengLi, TaoCui, JieYing, ZuobinCheng, JiujunAnhui Univ
Inst Phys Sci & Informat Technol Key Lab Intelligent Comp & Signal Proc Minist Educ Hefei 230039 Peoples R China Xidian Univ
Sch Cyber Engn Xian 710071 Peoples R China Anhui Univ
Sch Comp Sci & Technol Key Lab Intelligent Comp & Signal Proc Minist EducInst Phys Sci & Informat Technol Hefei 230039 Peoples R China Anhui Univ
Anhui Engn Lab IoT Secur Technol Hefei 230039 Peoples R China Tongji Univ
Key Lab Embedded Syst & Serv Comp Minist Educ Shanghai 200092 Peoples R China
Vehicles gather data collected by sensor nodes, combined with messages obtained from the other nodes in vehicular ad hoc networks (VANETs), to achieve safe driving. An announcement type of message is sent in the VANET...
详细信息
Vehicles gather data collected by sensor nodes, combined with messages obtained from the other nodes in vehicular ad hoc networks (VANETs), to achieve safe driving. An announcement type of message is sent in the VANET;it is collected by mobile vehicles, uploaded to a cloud server for storage, and provided to other vehicles for reference. However, in an open cloud environment, plaintext data are vulnerable to unauthorized access and even malicious tampering. To solve this issue, we propose an attribute-based encryption algorithm using blockchain, which is maintained by a roadside unit (RSU). The uploader's symmetric key is recorded on the blockchain, and all uploaded and accessed transactions are recorded for auditing. Our scheme can achieve the function of securely accessing different types of announcement messages according to different vehicle attributes. Security analysis and experimental results indicate that our scheme has achieved a balance between security and efficiency.
An expert-knowledge attention network (EKANet) was designed to improve the accuracy of arrhythmia diagnosis and reduce the recheck time. This network classifies four tachyarrhythmia on electrocardiogram (ECG) signals,...
详细信息
An expert-knowledge attention network (EKANet) was designed to improve the accuracy of arrhythmia diagnosis and reduce the recheck time. This network classifies four tachyarrhythmia on electrocardiogram (ECG) signals, encompassing most arrhythmia diseases. In the EKANet, two attention modules based on the knowledge of cardiology can rapidly capture the ECG rhythm and P waves in multiple leads without any training. This mechanism is performed to reduce the computational time of re-building a model. The EKANet integrates a six -layer convolutional neural network (CNN) and a gated recurrent unit (GRU) as the classifier to realise the tachyarrhythmia classification. The EKANet outperformed 1D CNN and ArrhythmiaNet on the MIT-BIH datasets by 3.1% on average accuracy. Furthermore, the EKANet achieved approximately 8.5% and 3.9% average F1 -score increases on the dataset of China ECG challenge contest compared with time-incremental CNN (TI-CNN) and attention-based TI-CNN, respectively. Meanwhile, the EKANet has a much lower complexity than that of the other typical models with a competitive accuracy.
In recent years, edge computing, as an important pillar for future networks, has been developing rapidly. Task offloading is a key part of edge computing that can provide computing resources for resource-constrained d...
详细信息
In recent years, edge computing, as an important pillar for future networks, has been developing rapidly. Task offloading is a key part of edge computing that can provide computing resources for resource-constrained devices to run computing-intensive applications, accelerate computing speed, and save energy. An efficient and feasible task offloading scheme can not only greatly improve the Quality of Experience (QoE) but also provide strong support and assistance for 5G/B5G networks, the Industrial Internet of Things (IIoT), computing networks, etc. To achieve these goals, this article proposes an adaptive edge task offloading scheme assisted by service deployment (SD-AETO) focusing on optimizing the energy utilization ratio (EUR) and the processing latency. In the preimplementation stage of the SD-AETO scheme, a service deployment scheme is invoked to assist with task offloading considering each service's popularity. The optimal service deployment scheme is obtained by using the approximate deployment graph (AD-graph). Furthermore, a task scheduling and queue offloading design procedure is proposed to complete the SD-AETO scheme based on task priority. The task priority is generated by corresponding service popularity and task offloading. Finally, we analyze our SD-AETO scheme and compare it with related approaches, and the results show that our scheme has a higher edge offloading rate and lower resource consumption for massive task scenarios in the edge network.
Target speaker information can be utilized in speech enhancement (SE) models to more effectively extract the desired speech. Previous works introduce the speaker embedding into speech enhancement models by means of co...
详细信息
Human Action Recognition is one of the most applied research directions in the field of Computer Vision, which is widely used in human-computer interaction, Augmented Reality (AR) technology, security monitoring, and ...
详细信息
ISBN:
(数字)9798350350920
ISBN:
(纸本)9798350350937
Human Action Recognition is one of the most applied research directions in the field of Computer Vision, which is widely used in human-computer interaction, Augmented Reality (AR) technology, security monitoring, and other scenarios. However, due to the complexity of human action gestures, existing Human Action Recognition methods have certain deficiencies in dealing with variable human gestures and action information, and the accuracy needs to be improved. To improve the accuracy, We propose a multi-dimensional network model based on SC-LSTM(Skip-Connection + LSTM). First, a Temporal Feature Extraction Module is designed based on SC-LSTM, and a Spatial Feature Extraction Module is designed based on CNN and Multi-Attention Mechanism to extract potential human action features from both temporal and spatial dimensions, respectively. Then, a separate SC-LSTM classification network is utilized to process these spatio-temporal features to obtain the final HAR results. The experimental results show that compared to other algorithms, the present model can more fully utilize the information in the temporal dimension, and thus performs better in terms of HAR accuracy.
暂无评论