检索结果-内蒙古大学图书馆

System-Level Energy-Aware Design Methodology Towards End-To-End Response Time Optimization

IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems 2019年

作者： Xie, Guoqi Huang, Jing Li, Yan Liu Renfa Li, Keqin Key Laboratory for Embedded and Network Computing of Hunan Province College of Computer Science and Electronic Engineering Hunan University Changsha Hunan 410082 China. College of Computer Science and Electronic Engineering Hunan University China 410082 China and also is with the Department of Computer Science State University of New York New Paltz New York 12561 USA.

Energy provision is known and limited for battery-constrained/powered systems (e.g., mobile phones, electric cars, robots, and wireless sensor networks). System-level energy-aware design methodologies have been proposed to control energy usage for energy provision guarantee, and to optimize the end-to-end response time for high-performance in heterogeneous energy-aware systems, where energy pre-assignment strategies are considered as their main contributions. However, when pre-assigning energy to tasks, these strategies either are overly pessimistic or adopt an average manner without a customizable approach to specific tasks. In this paper, we propose an effective algorithm called ratio-based energy pre-assignment (REP) to be customizable to split the available energy according to the energy ratios of individual tasks. Experiments confirm that REP outperforms the state-of-the-art three algorithms of the same type. After REP is used, we further propose a new innovation algorithm called energy surplus reduction (ESR) to reduce the execution time of small partial tasks, thereby optimizing the end-to-end response time within its energy provision. Experiments confirm ESR can reduce the energy usages of small partial tasks to achieve sufficient response time optimization. IEEE

关键词： Wireless sensor networks

来源：评论

学校读者我要写书评

暂无评论

A high-performance CNN method for offline handwritten Chinese character recognition and visualization

arXiv

引用

arXiv 2018年

作者： Melnyk, Pavlo You, Zhiqiang Li, Keqin Key Laboratory for Embedded and Network Computing of Hunan Province Hunan University Changsha410082 China College of Computer Science and Electronic Engineering Hunan University Changsha410082 China Department of Computer Science State University of New York at New Paltz New PaltzNY12561 United States

Recent researches introduced fast, compact and efficient convolutional neural networks (CNNs) for offline handwritten Chinese character recognition (HCCR). However, many of them did not address the problem of network interpretability. We propose a new architecture of a deep CNN with high recognition performance which is capable of learning deep features for visualization. A special characteristic of our model is the bottleneck layers which enable us to retain its expressiveness while reducing the number of multiply-accumulate operations and the required storage. We introduce a modification of global weighted average pooling (GWAP) - global weighted output average pooling (GWOAP). This paper demonstrates how they allow us to calculate class activation maps (CAMs) in order to indicate the most relevant input character image regions used by our CNN to identify a certain class. Evaluating on the ICDAR-2013 offline HCCR competition dataset, we show that our model enables a relative 0.83% error reduction while having 49% fewer parameters and the same computational cost compared to the current state-of-the-art single-network method trained only on handwritten data. Our solution outperforms even recent residual learning approaches. Copyright © 2018, The Authors. All rights reserved.

关键词： Visualization

来源：评论

学校读者我要写书评

暂无评论

Joint Partial Offloading and Resource Allocation for Distributed Multi-access Edge computing Enabled Vehicular network

TechRxiv

引用

TechRxiv 2022年

作者： Ma, Guifu Hu, Manjiang Wang, Xiaowei Wu, Di Li, Haoran Bian, Yougang Zhu, Konglin The State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body The College of Mechanical and Vehicle Engineering Hunan University Changsha410082 China Hunan University Wuxi214072 China The College of Design Hunan University Changsha410082 China The Key Laboratory for Embedded and Network Computing of Hunan Province Hunan University Changsha410082 China The College of Information and Communication Engineering Beijing University of Posts and Telecommunications Beijing100876 China

In the foreseeable Intelligent Transportation System (ITS), Intelligent Connected Vehicles (ICVs) will play an important role in improving travel efficiency and safety. However, it is challenging for ICVs to support the resource-hungry autonomous driving applications due to the limitation of hardware computing power. Fortunately, the emergence of Multi-access Edge computing (MEC) helps overcome this limitation effectively. In this paper, we investigate the vehicle and edge server collaborative computation offloading problem by jointly optimizing the partial offloading ratio and resource allocation ratio. Different from existing two-level centralized solution methods, we give a new perspective to deal with multi-variable non-convex optimization problem in a distributed manner. That is, by analyzing the feasibility conditions of the problem, we separate the coupling between variables and derive the bounds of partial offloading ratio and resource allocation ratio based on the constraints. We further separate the coupling variables among ICVs and transform the original problem into an equivalent form and decompose it into a two-level convex problem, where each sub-problem satisfies the task offloading requirements. We then develop two low-complexity distributed algorithms to obtain the optimal solution. Finally, extensive simulations demonstrate that our proposed algorithms can converge faster and significantly outperform the baseline schemes in terms of system cost. © 2022, CC BY.

关键词： Resource allocation

来源：评论

学校读者我要写书评

暂无评论

Identification of mild cognitive impairment using extreme learning machines model 11th

Identification of mild cognitive impairment using extreme le...

引用

11th International Conference on Intelligent computing, ICIC 2015

作者： Zhang, Wen Shen, Hao Ji, Zhiwei Meng, Guanmin Wang, Bing School of Electronics and Information Engineering Tongji University Shanghai201804 China The Advanced Research Institute of Intelligent Sensing Network Tongji University Shanghai201804 China The Key Laboratory of Embedded System and Service Computing Tongji University Shanghai201804 China Department of Clinical Laboratory Tongde Hospital of Zhejiang Province 234th Gucui Road Hangzhou310012 China

ISBN: (纸本)9783319221854

Alzheimer’s disease (AD) is very common disorder among the older people. Predicting mild cognitive impairment (MCI), an intermediate stage between normal cognition and dementia, in individuals with some symptoms of cognitive decline may have great influence on treatment choice and disease progression. In this work, a novel prediction model based on extreme learning machines algorithm was proposed to identify mild cognitive impairment using the information of some biomarkers, i.e., Position Emission Tomography (PET), Magnetic Resonance Imaging (MRI), and the cerebrospinal fluid (CSF). Firstly, each subject will be represented by three biomarkers using a vector with 201 dimensions after some essential pre-processing. Then, a prediction model based on extreme learning machine algorithm was trained to catch the difference between the patients and healthy persons. Finally, the model had been employed to identify identifying MCI participants from the normal people. Based on a dataset, including 99 MCI patients and 52 health controls, our proposed method achieved 70.86 % prediction accuracy with 67.68 % sensitivity at the precision of 76.92 %. Extensive experiments are performed to compare our method with state-of-the-art techniques, i.e., support vector machine (SVM) and SVM with fusion kernels. Experimental results demonstrate that proposed extreme learning machine is a powerful tool for predicting MCI with excellent performance and less time. © Springer International Publishing Switzerland 2015.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Efficient poisoning attacks and defenses for unlabeled data in DDoS prediction of intelligent transportation systems

引用

Security and Safety 2022年第1期1卷 121-141页

作者： Zhong Li Xianke Wu Changjun Jiang College of Information Science and Technology Donghua University Key Laboratory of the Ministry of Education for Embedded System and Service Computing Department of Computer Science Tongji University Shanghai Network Financial Security Collaborative Innovation Center Tongji University

Nowadays, large numbers of smart sensors(e.g., road-side cameras) which communicate with nearby base stations could launch distributed denial of services(DDoS) attack storms in intelligent transportation systems. DDoS attacks disable the services provided by base stations. Thus in this paper, considering the uneven communication traffic flows and privacy preserving, we give a hidden Markov model-based prediction model by utilizing the multi-step characteristic of DDoS with a federated learning framework to predict whether DDoS attacks will happen on base stations in the future. However, in the federated learning,we need to consider the problem of poisoning attacks due to malicious participants. The poisoning attacks will lead to the intelligent transportation systems paralysis without security protection. Traditional poisoning attacks mainly apply to the classification model with labeled data. In this paper, we propose a reinforcement learning-based poisoning method specifically for poisoning the prediction model with unlabeled data. Besides, previous related defense strategies rely on validation datasets with labeled data in the server. However, it is unrealistic since the local training datasets are not uploaded to the server due to privacy preserving, and our datasets are also unlabeled. Furthermore, we give a validation dataset-free defense strategy based on Dempster–Shafer(D–S) evidence theory avoiding anomaly aggregation to obtain a robust global model for precise DDoS prediction. In our experiments, we simulate 3000 points in combination with DARPA2000 dataset to carry out evaluations. The results indicate that our poisoning method can successfully poison the global prediction model with unlabeled data in a short time. Meanwhile, we compare our proposed defense algorithm with three popularly used defense algorithms. The results show that our defense method has a high accuracy rate of excluding poisoners and can obtain a high attack prediction probability.

关键词： Poisoning attacks Defenses Multi-step DDoS prediction Unlabeled data Intelligent transportation systems

来源：评论

学校读者我要写书评

暂无评论

Implicit privacy preservation:a framework based on data generation

引用

Security and Safety 2022年第1期1卷 166-183页

作者： Qing Yang Cheng Wang Teng Hu Xue Chen Changjun Jiang Key Laboratory of Embedded System and Service Computing(Tongji University) Ministry of EducationShanghai 201804China National(Province-Ministry Joint)Collaborative Innovation Center for Financial Network Security Tongji UniversityShanghai 201804China

This paper addresses a special and imperceptible class of privacy,called implicit *** contrast to traditional(explicit)privacy,implicit privacy has two essential properties:(1)It is not initially de ned as a privacy attribute;(2)it is strongly associated with privacy *** other words,attackers could utilize it to infer privacy attributes with a certain probability,indirectly resulting in the disclosure of private *** deal with the implicit privacy disclosure problem,we give a measurable de nition of implicit privacy,and propose an ex-ante implicit privacy-preserving framework based on data generation,called *** framework consists of an implicit privacy detection module and an implicit privacy protection *** former uses normalized mutual information to detect implicit privacy attributes that are strongly related to traditional privacy *** on the idea of data generation,the latter equips the Generative Adversarial network(GAN)framework with an additional discriminator,which is used to eliminate the association between traditional privacy attributes and implicit *** elaborate a theoretical analysis for the convergence of the *** demonstrate that with the learned generator,IMPOSTER can alleviate the disclosure of implicit privacy while maintaining good data utility.

关键词： Privacy preservation Implicit privacy Generative adversarial network Data utility Data generation

来源：评论

学校读者我要写书评

暂无评论

DCOM-GNN: A Deep Clustering Optimization Method for Graph

SSRN

引用

SSRN 2023年

作者： Yang, Haoran Wang, Junli Duan, Rui Yan, Chungang Key Laboratory of Embedded System and Service Computing Ministry of Education Shanghai201804 China Collaborative Innovation Center for Financial Network Security Tongji University Shanghai201804 China

Deep clustering plays an important role in data analysis, and with the prevalence of graph data nowadays, various deep clustering models on graph are constantly proposed. However, due to the lack of more adequate clustering guidance, the discriminability of feature representation learned from these models for the clustering task is limited. Therefore, for the purpose of enabling the output of these models to be more cluster-oriented, we propose a Deep Clustering Optimization Method for Graph Neural networks (DCOM-GNN), which can be attached to the original model architecture conveniently. For DCOM-GNN, it contains two components, one is the inter-cluster distance optimization module, whose role is to further adjust the distance between clusters of the original model output rationally. Another one is the intra-cluster distance optimization module, which aims to improve the cohesiveness of the original model output. Comprehensive experiments show that the performance of various deep clustering models on graph can be significantly improved after adding DCOM-GNN. © 2023, The Authors. All rights reserved.

关键词： Cluster analysis

来源：评论

学校读者我要写书评

暂无评论

A rich ranking model based on the matthew effect optimization 7th

A rich ranking model based on the matthew effect optimizatio...

引用

7th International Conference on Computational Data and Social networks, CSoNet 2018

作者： Li, Jinzhong Liu, Guanjun Department of Computer Science and Technology College of Electronic and Information Engineering Jinggangshan University Ji’an343009 China Network and Data Security Key Laboratory of Sichuan Province University of Electronic Science and Technology of China Chengdu610054 China Department of Computer Science and Technology College of Electronic and Information Engineering Tongji University Shanghai201804 China Key Laboratory of Embedded System and Service Computing Ministry of Education Tongji University Shanghai201804 China

ISBN: (纸本)9783030046477

Most existing approaches of learning to rank treat the effectiveness of each query equally which results in a relatively lower ratio of queries with high effectiveness (i.e. rich queries) in the produced ranking model. Such ranking models need to be further optimized to increase the number of rich queries. In this paper, queries with different effectiveness are distinguished, and the queries with higher effectiveness are given higher weights. We modify the gradient in the LambdaMART algorithm based on a new perspective of Matthew effect to highlight the optimization of the rich queries and to produce the rich ranking model, and we present a consistency theorem for the modified optimization objective. Based on the effectiveness evaluation criteria for information retrieval, we introduce the Gini coefficient, mean-variance and quantity statistics to measure the performances of the ranking models. Experimental results show that the ranking models produced by the gradient-modified LambdaMART algorithm based on Matthew effect exhibit a stronger Matthew effect compared to the original LambdaMART algorithm. © Springer Nature Switzerland AG 2018.

关键词： Artificial intelligence

来源：评论

学校读者我要写书评

暂无评论

A Multi-label Classifier for Human Protein Subcellular Localization Based on LSTM networks

A Multi-label Classifier for Human Protein Subcellular Local...

引用

2018 International Conference on Advanced Control, Automation and Artificial Intelligence (ACAAI2018)

作者： Zhiying Gao Lijun Sun Zhihua Wei Department of Computer Science and Technology Tongji University Research Center of Big Data and Network Security Tongji University Key Laboratory of Embedded System and Service Computing Tongji University

Nowadays, with the increasing number of protein sequences all over the world, more and more people are paying their attention to predicting protein subcellular location. Since wet experiment is costly and time-consuming, the automatic computational methods are urgent. In this paper, we propose a variant model based on Long Short-Term Memory(LSTM) to predict protein subcellular location. In this model, we employ LSTM to capture long distance dependency features of the sequence data. Moreover, we adjust the loss function of the loss layer to solve multi-label classification problem. Experimental results demonstrate that, compared with the traditional machine learning methods, our method achieves the best performance in various evaluation metrics.

关键词： LSTM multi-label classification protein subcellular localization

来源：评论

学校读者我要写书评

暂无评论

Graph Ensemble Neural network

SSRN

引用

SSRN 2023年

作者： Duan, Rui Yan, Chungang Wang, Junli Jiang, Changjun Key Laboratory of Embedded System and Service Computing Ministry of Education Shanghai201804 China Collaborative Innovation Center for Financial Network Security Tongji University Shanghai201804 China

Ensemble methods have been shown to improve graph neural networks (GNNs). Existing ensemble methods on graphs determine a strong classifier by combining a set of trained base classifiers, i.e., combining the final outputs of base classifiers for prediction. However, these methods fail to promote many popular GNNs to perform well under heterophily (in graphs where many connected nodes havedifferent class labels), which limits their applicability. Furthermore, they ignore the hierarchical nature of GNNs, which results in no interaction between base classifiers when neighbors are aggregated (during training). Two issues arise from this: low applicability and shallow ensemble. We propose Graph Ensemble Neural network (GEN) for addressing above issues, which is not a simple ensembleof GNNs, but instead integrates ensemble into GNNs to fuse a set of graphs. GEN deepens single ensemble into multiple ensembles during training and applies to homophily and heterophily graphs. In GEN, we design structure augmentation to generate some graphs for training and design feature augmentation for attenuating errors brought by the initial features. Different from existing graphensemble methods that execute only one ensemble, GEN executes multiple deep ensembles throughout the neighbor aggregation to fuse multiple graphs generated by structure augmentation. Extensive experiments show that GEN achieves new state-of-the-art performance on homophily and heterophily graphs for the semi- and full-supervised node classification. The source code of GEN is publicly available at https://***/graphNN/GEN1. © 2023, The Authors. All rights reserved.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：