检索结果-内蒙古大学图书馆

Optimizing Word Embedding for Fine-Grained Sentiment Analysis 1

5th International Conference on Artificial Intelligence and Security, ICAIS 2019

作者： Zhang, Wei Zhang, Yue Yang, Kehua Key Laboratory for Embedded and Network Computing of Hunan Province College of Computer Science and Electronic Engineering Hunan University Changsha410082 China

ISBN: (数字)9783030242657

ISBN: (纸本)9783030242640

Word embeddings have been extensively used for various Natural Language Processing tasks. However, word vectors trained based on corpus context information fail to distinguish words with the same context but different semantics, which may lead to a similar word vector with opposite semantic terms. This will affect some Natural Language Processing tasks, such as fine-grained sentiment analysis tasks. In this paper, a new word vectors optimization model is proposed. This model can be applied to any pre-trained word vectors. Within a certain range, it can make the opposite semantic words away from each other, and the same semantic words are close to each other. The experimental results show that our model can improve the traditional word embedding in the fine-grained emotional analysis task of Chinese Weibo, and the optimized word vector using our model outperforms the unoptimized word vector. © 2019, Springer Nature Switzerland AG.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

Security-aware Signal Packing Algorithm for CAN-based Automotive Cyber-physical Systems

引用

IEEE/CAA Journal of Automatica Sinica 2015年第4期2卷 422-430页

作者： Yong Xie Liangjiao Liu Renfa Li Jianqiang Hu Yong Han Xin Peng Key Laboratory for Application Technology of Internet of Things of Fujian Educational Bureau Xiamen University of Technology Key Laboratory of Embedded and Networking Computing of Hunan Province Hunan University Key Laboratory on Complex Systems Optimization & Controlling of Hunan High Education Institutions Hunan Institute of Science and Technology

network and software integration pose severe challenges in cyber-security for controller area network (CAN)-based automotive cyber-physical system (ACPS), therefore we employ message authentication code (MAC) to defend CAN against masquerade attack, but the consequent bandwidth overhead makes it a necessity to find the tradeoff among security, real-time and bandwidth utilization for signal packing problem (SPP) of CAN. A mixed-security signal model is firstly proposed to formally describe the properties and requirements on security and real-time for signals, and then a mixed-integer linear programming (MILP) formulation of SPP security-aware signal packing (SASP) is implemented to solve the tradeoff problem, where the bandwidth utilization is improved and the requirements in both security and real-time are met. Experiments based on both society of automotive engineers (SAE) standard signal set and simulated signal set showed the effectiveness of SASP by comparing with the state-of-the-art algorithm. © 2014 Chinese Association of Automation.

关键词： Algorithms Bandwidth Communication channels (information theory) Control system synthesis embedded systems Integer programming

来源：评论

学校读者我要写书评

暂无评论

Real-time depth estimation with an optimized encoder-decoder architecture on embedded devices 21

Real-time depth estimation with an optimized encoder-decoder...

引用

21st IEEE International Conference on High Performance computing and Communications, 17th IEEE International Conference on Smart City and 5th IEEE International Conference on Data Science and Systems, HPCC/SmartCity/DSS 2019

作者： Tu, Xiaohan Xu, Cheng Liu, Siping Xie, Guoqi Li, Renfa Key Laboratory for Embedded and Network Computing of Hunan Province China College of Computer Science and Electronic Engineering Hunan University Changsha410082 China

ISBN: (纸本)9781728120584

As a strategy to obtain dense depth maps from a single image, sparse depths, or the fusion of both (RGBd), depth estimation receives much attention. Usually relying on high-performance workstations, existing depth prediction methods, are not particularly suitable for embedded devices. Moreover, the accuracy of depth estimation needs to be improved. Towards real-time and reliable depth prediction on portable embedded platforms, we propose an encoder-decoder architecture (EDA). Specifically, these are the three contributions of this work. (1) We present a light-weight encoder, which takes three modalities of data as inputs, namely, RGB images, sparse depths, or RGBd. The encoder is efficient to extract features on portable devices. (2) A novel decoder is designed to obtain accurate and high-resolution depth maps. (3) We use the existing TVM to deploy, compile, and optimize EDA on embedded platforms, for further improvements in the computational speed and power of depth estimation. In real scenarios, the optimized EDA achieves remarkable results with low latency. Through extensive experimental evaluation on two public datasets, we show that the optimized EDA outperforms the state of the arts in accuracy and computational resources. To the best of our knowledge, this is the first paper that provides an efficient encoder, learnable decoder, and end-to-end optimized encoder-decoder architecture for precise depth prediction. © 2019 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Locally differentially private high-dimensional data synthesis

引用

Science China(Information Sciences) 2023年第1期66卷 25-42页

作者： Xue CHEN Cheng WANG Qing YANG Teng HU Changjun JIANG Key Laboratory of Embedded System and Service Computing (Tongji University) Ministry of Education National (Province-Ministry Joint) Collaborative Innovation Center for Financial Network Security Tongji University

In local differential privacy(LDP), a challenging problem is the ability to generate highdimensional data while efficiently capturing the correlation between attributes in a dataset. Existing solutions for low-dimensional data synthesis, which partition the privacy budget among all attributes, cease to be effective in high-dimensional scenarios due to the large-scale noise and communication cost caused by the high dimension. In fact, the high-dimensional characteristics not only bring challenges but also make it possible to apply some technologies to break this bottleneck. This paper presents Sam Priv Syn for high-dimensional data synthesis under LDP, which is composed of a marginal sampling module and a data generation *** marginal sampling module is used to sample from the original data to obtain two-way marginals. The sampling process is based on mutual information, which is updated iteratively to retain, as much as possible,the correlation between attributes. The data generation module is used to reconstruct the synthetic dataset from the sampled two-way marginals. Furthermore, this study conducted comparison experiments on the real-world datasets to demonstrate the effectiveness and efficiency of the proposed method, with results proving that Sam PrivSyn can not only protect privacy but also retain the correlation information between the attributes.

关键词： local differential privacy high-dimensional data synthesis sampling data privacy privacy-preserving protocols

来源：评论

学校读者我要写书评

暂无评论

A Novel Neighborhood-Weighted Sampling Method for Imbalanced Datasets

引用

Chinese Journal of Electronics 2022年第5期31卷 969-979页

作者： GUANG Mingjian YAN Chungang LIU Guanjun WANG Junli JIANG Changjun Key Laboratory of Embedded System and Service Computing Tongji University Ministry of Education National (Province-Ministry Joint) Collaborative Innovation Center for Financial Network Security Tongji University

The weighted sampling methods based on k-nearest neighbors have been demonstrated to be effective in solving the class imbalance problem. However,they usually ignore the positional relationship between a sample and the heterogeneous samples in its neighborhood when calculating sample weight. This paper proposes a novel neighborhood-weighted based sampling method named NWBBagging to improve the Bagging algorithm's performance on imbalanced datasets. It considers the positional relationship between the center sample and the heterogeneous samples in its neighborhood when identifying critical samples. And a parameter reduction method is proposed and combined into the ensemble learning framework, which reduces the parameters and increases the classifier's diversity. We compare NWBBagging with some state-of-the-art ensemble learning algorithms on 34 imbalanced datasets, and the result shows that NWBBagging achieves better performance.

关键词： parameter reduction method neighborhood-weighted sampling method pattern classification class imbalance problem NWBBagging k-nearest neighbors sampling methods imbalanced datasets nearest neighbour methods ensemble learning algorithm Bagging algorithm

来源：评论

学校读者我要写书评

暂无评论

TMI: Two-dimensional maintainability index for automotive software maintainability measurement

引用

Journal of Systems Architecture 2025年 166卷

作者： Jiao Xie Jialing Yang Sirong Zhao Jianmei Lei Kaiwei Fan Guoqi Xie Key Laboratory for Embedded and Network Computing of Hunan Province College of Computer Science and Electronic Engineering Hunan University Changsha 410082 Hunan China Center of Convergence of Automobile and Cyberspace Research Institute of Hunan University in Chongqing Chongqing 401120 China China Automotive Engineering Research Institute Co Ltd Electronic Communication and Software Evaluation Research Center Chongqing 401122 China

The mass adoption of autonomous driving and Advanced Driver-Assistance System (ADAS) technologies increases the complexity and maintenance cost of automotive-embedded software during model-based development. Through model-based development tools (e.g., MATLAB/Simulink), the code is automatically generated by architectural transformation; however, unreasonable software architecture design could make the final generated code be difficult to maintain. Currently, both Microsoft Maintainability Index (MMI) and Bosch Maintainability Index (BMI) are used to measure the maintainability of code, but neither software architecture. We establish a four-layer maintainability index model by analyzing the characteristics of automotive software architecture, software quality model, and industry-related standards. We then propose a Two-dimensional Maintainability Index (TMI), meaning the code and architecture dimensions are simultaneously considered. This is the first work of maintainability measurement to combine code and architecture dimensions while ensuring low redundancy. TMI utilizes nine index metrics to guide architecture refactoring and code optimization, significantly improving software maintainability.

关键词： MATLAB

来源：评论

学校读者我要写书评

暂无评论

FPGA based optimized SHA-3 finalist in reconfigurable hardware

FPGA based optimized SHA-3 finalist in reconfigurable hardwa...

引用

2011 International Symposium on Integrated Circuits, ISIC 2011

作者： Song, Qian Wang, Yi Li, Zhican Zhou, Quan Wu, Wufei Han, Demin Xu, Wenlong Chen, Zuo Li, Renfa Embedded Systems and Networking Laboratory Hunan University Hunan Provincial Key Laboratory of Network and Information Security China

ISBN: (纸本)9781612848648

A hash function is well-defined procedure to convert large, uncertain long message into fixed small integers. Secure Hash Algorithm (SHA) is an one-way message digest algorithm which is usually used in cryptographic applications such as authentication, digital signature and data integrity. In this paper, we proposed the reconfigurable structure for SHA-3 finalist BLAKE, Grøstl, JH, Keccak and Skein, separately. The proposed reconfigurable Grøstl, JH and Keccak could support different digested sizes. And Skein and BLAKE optimized three different modes using one single hardware core. The experimental results showed that our proposed structure could support different parameters of SHA-3 finalist with comparable performance among the existing works when ported to Xilinx Virtex-5 FPGA platform. © 2011 IEEE.

关键词： Field programmable gate arrays (FPGA)

来源：评论

学校读者我要写书评

暂无评论

Instance-Aware Style-Swap for Disentangled Attribute-Level Image Editing 17th

Instance-Aware Style-Swap for Disentangled Attribute-Level ...

引用

17th CCF Conference on Computer Supported Cooperative Work and Social computing, ChineseCSCW 2022

作者： Zhou, Xinjiao Jiang, Bin Yang, Chao Hu, Haotian Sun, Minyu College of Computer Science and Electronic Engineering Hunan University Hunan Changsha410082 China Key Laboratory for Embedded and Network Computing of Hunan Province Hunan University Hunan Changsha410082 China

ISBN: (纸本)9789819923847

Recent studies have shown that attribute-level image editing could be achieved by modifying the latent code of style-based Generative Adversarial network (StyleGAN). However, many existing methods suffer from semantic entanglement, which leads to undesirable changes when editing a specific attribute. To solve this problem, we focus on the S space of StyleGAN2 and propose a simple and effective Optimization-based Instance-Aware Style-Swap method, which could perform a natural transformation effect on arbitrary source images and reference images by searching instance-aware swap coefficients with the guidance of prior information. By further analyzing the learned swapping coefficients, we can identify the style channels connected with a specific target attribute. We verify the effectiveness of our proposed methods in the field of face attribute editing. Extensive experiments have demonstrated that our method can achieve controllable and fine-grained image editing along various attributes. Qualitative and quantitative results show the advantages of our method compared with other semantic image editing methods. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Collaborative Multi-head Contextualized Sparse Representations for Real-Time Open-Domain Question Answering 17th

Collaborative Multi-head Contextualized Sparse Representatio...

引用

17th CCF Conference on Computer Supported Cooperative Work and Social computing, ChineseCSCW 2022

作者： Sun, Minyu Jiang, Bin Zhou, Xinjiao Zhang, Bolin Yang, Chao College of Computer Science and Electronic Engineering Hunan University Hunan Changsha410082 China Key Laboratory for Embedded and Network Computing of Hunan Province Hunan University Hunan Changsha410082 China

ISBN: (纸本)9789819923847

An efficient method of representing and retrieving information is an essential component of open domain QA. There are question and answer models that allow for real-time responses with speed benefit and scalability. Nonetheless, due to the limitations of existing phrase models, their accuracy is low. In this paper, we improve the contextualized sparse representation to strengthen the connection between contextual information. We achieve better answer retrieval by enhancing the embedding quality of the model for phrase representation. Specifically, based on original contextualized sparse representations, we transform the single self-attention into collaborative multi-head attention so that attention heads can connect and pay attention to crucial information in different context locations. Compared with learning sparse vectors in n-gram vocabulary space by rectified self-attention, collaborative multi-head attention performs better on the SQuAD dataset. Due to the increased efficiency of critical information representation, the model improves to varying degrees on both of the two evaluation metrics. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Vector spaces

来源：评论

学校读者我要写书评

暂无评论

Secret sharing based countermeasure for AES S-box

Secret sharing based countermeasure for AES S-box

引用

2011 International Symposium on Integrated Circuits, ISIC 2011

作者： Wang, Yi Yuan, Zheng Li, Zhican Li, Renfa Embedded Systems and Networking Laboratory Hunan University Hunan Provincial Key Laboratory of Network and Information Security Changsha China

ISBN: (纸本)9781612848648

Cryptographic devices are vulnerable to Differential Power Attack (DPA) in embedded systems. Masking methods are popularly used to defend against DPA by masking all intermediate data with random values. However, masking schemes on algorithm level are vulnerable to Higher-Order DPA (HODPA), while on gate level glitch attack is the biggest threaten attack. In this paper, we proposed a secret sharing based countermeasure for AES S-box, which can defend against both HODPA and glitch attack. The experimental results show that our proposed design takes up less hardware resources and achieves faster speed compared with the existing methods. © 2011 IEEE.

关键词： Field programmable gate arrays (FPGA)

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：