检索结果-内蒙古大学图书馆

arXiv 2020年

作者： Li, Ming Sonoda, Sho Cao, Feilong Wang, Yu Guang Liang, Jiye Key Laboratory of Intelligent Education Technology Application of Zhejiang Province Zhejiang Normal University Jinhua China Deep Learning Theory Team RIKEN AIP Tokyo Japan College of Sciences China Jiliang University Hangzhou China Institute of Natural Sciences Shanghai Jiao Tong University Shanghai China School of Mathematical Sciences Shanghai Jiao Tong University Shanghai China Key Laboratory of Computational Intelligence and Chinese Information Processing of Ministry of Education School of Computer and Information Technology Shanxi University Taiyuan China

We investigate the expressive power of depth-2 bandlimited random neural networks. A random net is a neural network where the hidden layer parameters are frozen with random assignment, and only the output layer parameters are trained by loss minimization. Using random weights for a hidden layer is an effective method to avoid non-convex optimization in standard gradient descent learning. It has also been adopted in recent deep learning theories. Despite the well-known fact that a neural network is a universal approximator, in this study, we mathematically show that when hidden parameters are distributed in a bounded domain, the network may not achieve zero approximation error. In particular, we derive a new nontrivial approximation error lower bound. The proof utilizes the technique of ridgelet analysis, a harmonic analysis method designed for neural networks. This method is inspired by fundamental principles in classical signal processing, specifically the idea that signals with limited bandwidth may not always be able to perfectly reconstruct the original signal. We corroborate our theoretical results with various simulation studies, and generally, two main take-home messages are offered: (i) Not any distribution for selecting random weights is feasible to build a universal approximator;(ii) A suitable assignment of random weights exists but to some degree is associated with the complexity of the target function. © 2020, CC BY.

关键词： Convex optimization

来源：评论

学校读者我要写书评

暂无评论

A Research on the Control System of High-Speed Homopolar Motor with Solid Rotor Based on Flywheel Energy Storage

引用

Complexity 2020年 2020卷

作者： Jing, Lili Yu, Yandong Xue, Xiaochuan Key Laboratory of High Speed Signal Processing and Internet of Things Technology Application Jining Normal University Ulanqab Inner Mongolia 012000 China Department of Computer Science Jining Normal University Ulanqab Inner Mongolia 012000 China Systems Engineering Research Institute of China State Shipbuilding Corporation Beijing 100191 China

In view of the defects of the motors used for flywheel energy storage such as great iron loss in rotation, poor rotor strength, and robustness, a new type of motor called electrically excited homopolar motor is adopted in this paper for flywheel energy storage. Compared to general motors, this motor has the advantages of simple structure, high rotor strength, and low iron loss in rotation. A double closed-loop PI governing system of the new motor was designed, modeled, and simulated with this motor as the controlled object on simulation platform. The simulation result shows that the PI-controlled electrically excited homopolar motor system realized fast speed governing and good system performance, achieved the purpose of quick charge by flywheel energy storage, verified the correctness and effectiveness of the design scheme, and provided data reference and theoretical basis for the practice of future flywheel energy storage experiments. © 2020 Lili Jing et al.

关键词： Digital storage

来源：评论

学校读者我要写书评

暂无评论

QuesNet: A unified representation for heterogeneous test questions

arXiv

引用

arXiv 2019年

作者： Yin, Yu Liu, Qi Huang, Zhenya Chen, Enhong Tong, Wei Wang, Shijin Su, Yu Anhui Province Key Laboratory of Big Data Analysis and Application School of Computer Science and Technology School of Data Science University of Science and Technology of China IFLYTEK Research IFLYTEK CO. LTD. State Key Laboratory of Cognitive Intelligence

Understanding learning materials (e.g. test questions) is a crucial issue in online learning systems, which can promote many applications in education domain. Unfortunately, many supervised approaches suffer from the problem of scarce human labeled data, whereas abundant unlabeled resources are highly underutilized. To alleviate this problem, an effective solution is to use pre-trained representations for question understanding. However, existing pretraining methods in NLP area are infeasible to learn test question representations due to several domain-specific characteristics in education. First, questions usually comprise of heterogeneous data including content text, images and side information. Second, there exists both basic linguistic information as well as domain logic and knowledge. To this end, in this paper, we propose a novel pre-training method, namely QuesNet, for comprehensively learning question representations. Specifically, we first design a unified framework to aggregate question information with its heterogeneous inputs into a comprehensive vector. Then we propose a two-level hierarchical pre-training algorithm to learn better understanding of test questions in an unsupervised way. Here, a novel holed language model objective is developed to extract low-level linguistic features, and a domain-oriented objective is proposed to learn high-level logic and knowledge. Moreover,we showthat Ques- Net has good capability of being fine-tuned in many question-based tasks. We conduct extensive experiments on large-scale real-world question data, where the experimental results clearly demonstrate the effectiveness of QuesNet for question understanding as well as its superior applicability. Copyright © 2019, The Authors. All rights reserved.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

Optical polarization imaging for underwater target detection with non-scatter background

引用

Journal of Measurement Science and Instrumentation 2020年第4期11卷 335-342页

作者： GUAN Jin-ge ZHAO Yong ZHENG Yong-qiu MA Miao SUN Peng XUE Chen-yang School of Information and Communication Engineering North University of ChinaTaiyuan 030051China Beijing Institute of Computer Application Technology China North Industries Group Corporation LimitedBeijing 100089China Key Laboratory of Instrumentation Science and Dynamic Measurement(North University of China) Ministry of EducationTaiyuan 030051China Department of Science Taiyuan Institute of TechnologyTaiyuan 030008China

For conventional optical polarization imaging of underwater target,the polarization degree of backscatter should be pre-measured by averaging the pixel intensities in the no target region of the polarization images,and the polarization property of the target is assumed to be completely *** the scattering background is unseen in the field of view or the target is polarized,conventional method is helpless in detecting the *** improvement is to use lots of co-polarization and cross polarization detection *** propose a polarization subtraction method to estimate depolarization property of the scattering noise and target *** experiment in a quartz cuvette container is performed to demonstrate the effectiveness of the proposed *** results show that the proposed method can work without scattering background reference,and further recover the target along with smooth surface for polarization preserving *** study promotes the development of optical polarization imaging systems in underwater environments.

关键词： polarization imaging underwater optical scattering optical information processing target detection

来源：评论

学校读者我要写书评

暂无评论

Nonlinear Cross-Domain Feature Representation Learning Method Based on Dual Constraints

Nonlinear Cross-Domain Feature Representation Learning Metho...

引用

IEEE International Conference on Big Knowledge (ICBK)

作者： Han Ding Yuhong Zhang Shuai Yang Yaojin Lin Key Laboratory of Knowledge Engineering with Big Data (Hefei University of Technology) Ministry of Education School of Computer Science and Information Engineering Hefei University of Technology Hefei China Key Laboratory of Data Science and Intelligence Application Fujian Province University Zhangzhou China

Feature representation learning is a research focus in domain adaptation. Recently, due to the fast training speed, the marginalized Denoising Autoencoder (mDA) as a standing deep learning model has been widely utilized for feature representation learning. However, the training of mDA suffers from the lack of nonlinear relationship and does not explicitly consider the distribution discrepancy between domains. To address these problems, this paper proposes a novel method for feature representation learning, namely Nonlinear cross-domain Feature learning based Dual Constraints (NFDC), which consists of kernelization and dual constraints. Firstly, we introduce kernelization to effectively extract nonlinear relationship in feature representation learning. Secondly, we design dual constraints including Maximum Mean Discrepancy (MMD) and Manifold Regularization (MR) in order to minimize distribution discrepancy during the training process. Experimental results show that our approach is superior to several state-of-the-art methods in domain adaptation tasks.

关键词： Noise reduction Manifolds Training Adaptation models Kernel Deep learning Feature extraction

来源：评论

学校读者我要写书评

暂无评论

FrepJoin:an efficient partition-based algorithm for edit similarity join

引用

Frontiers of Information technology & Electronic Engineering 2017年第10期18卷 1499-1510页

作者： Ji-zhou LUO Sheng-fei SHI Hong-zhi WANG Jian-zhong LI School of Computer Science and Technology Harbin Institute of Technology Guangdong Key Laboratory of Popular High Performance Computers Key Laboratory of Service Computing and Application

String similarity join(SSJ) is essential for many applications where near-duplicate objects need to be found. This paper targets SSJ with edit distance constraints. The existing algorithms usually adopt the filter-andrefine framework. They cannot catch the dissimilarity between string subsets, and do not fully exploit the statistics such as the frequencies of characters. We investigate to develop a partition-based algorithm by using such *** frequency vectors are used to partition datasets into data chunks with dissimilarity between them being caught easily. A novel algorithm is designed to accelerate SSJ via the partitioned data. A new filter is proposed to leverage the statistics to avoid computing edit distances for a noticeable proportion of candidate pairs which survive the existing filters. Our algorithm outperforms alternative methods notably on real datasets.

关键词： String similarity join Edit distance Filter and refine Data partition Combined frequency vectors

来源：评论

学校读者我要写书评

暂无评论

Research and Prospect of Digital Economy Development in Shandong Province

Research and Prospect of Digital Economy Development in Shan...

引用

2019 5th International Conference on Humanities and Social Science Research (ICHSSR 2019)

作者： Zhengqian Feng Hao Wang Liping Wei Shandong Information Resources Application Association Qilu University of Technology(Shandong Academy of Sciences) Shandong Computer Science Center(National Supercomputer Center in Jinan) Shandong Provincial Key Laboratory of Computer Networks Shandong University of Science and Technology

Digital economy refers to the economic model that takes digital technology as the core to drive the whole economic activity process and create *** the future, all economic links may be driven by digital technology, digital technology will be the main driving force for world economic development, and digital economy will be the new engine for world economic *** great transformation of digital economy is a new opportunity for economic *** development level of digital economy in Shandong province has entered the first echelon in China, but there are still some factors restricting its *** digital economy has become the core force leading the scientific and technological revolution and industrial transformation, and human society is entering a new stage marked by digital *** province must correctly grasp the general trend of the development of digital economy, do a good job in planning and creating momentum and taking advantage of it, accelerate the construction of a strong digital province, make the digital economy become an important starting point to promote the conversion of old and new drives and an important supporting force of "lead the way".

关键词： Digital Economy Development Status Evaluation Index

来源：评论

学校读者我要写书评

暂无评论

Single Image Super Resolution Based on Generative Adversarial Networks

Single Image Super Resolution Based on Generative Adversaria...

引用

2019第十一届数字图像处理国际会议

作者： Kai Li Ye Liang Shenghao Yang Jinfang Jia Jianqiang Huang Xiaoying Wang State Key Laboratory of Plateau Ecology and Agriculture Department of Computer Technology and ApplicationQinghai University Rocket Force University of Engineering Basic Course DepForeign language Office

Deep neural networks based on SRGAN single image super-resolution reconstruction can generate more realistic images than CNN-based super-resolution deep neural ***,when the network is deeper and more complex,unpleasant artifacts can *** a lot of experiments,we can use the ESRGAN model to avoid such *** using the ESRGAN model for super-resolution reconstruction,the perceived index of the resulting results does not reach a lower *** are two reasons for this:（1）ESRGAN does not expand the feature *** uses 128*128 to obtain the feature information of the image by default,and can’t get more image information better.（2） ESRGAN did not re-optimize the generated ***,we propose ESRGAN-Pro to optimize ESRGAN for the above two aspects,combined with a large amount of training data,and get a better perception index and texture.

关键词： Generative Adversarial Networks super resolution deep learning framework image processing

来源：评论

学校读者我要写书评

暂无评论

An embedding-based approach to constructing OWL ontologies

An embedding-based approach to constructing OWL ontologies

引用

2018 ISWC Posters and Demonstrations, Industry and Blue Sky Ideas Tracks, ISWC-P and D-Industry-BlueSky 2018

作者： Zhang, Lijing Zhang, Xiaowang Zhao, Leyuan Tian, Jiachen Chen, Shizhan Wu, Hong Wang, Kewen Feng, Zhiyong School of Computer Science and Technology Tianjin University Tianjin China School of Computer Software Tianjin University Tianjin China Tianjin Key Laboratory of Cognitive Computing and Application Tianjin China

This paper presents a novel system OWLearner for automatically extracting axioms for OWL ontologies from RDF data using embedding models. In this system, ontology construction is transformed to the classification problem in machine learning and thus off-the-shelf tools can be employed to learn axioms in OWL. There are mainly three modules, namely, embedding, sampling, and training & learning. Large ontologies DBpedia and YAGO are used to validate the proposed approach. The experimental results show that OWLearner is able to learn high-quality expressive OWL axioms automatically and efficiently. © 2018 CEUR-WS. All rights reserved.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

An ontology-based approach to adaptive data processing

An ontology-based approach to adaptive data processing

引用

2018 ISWC Posters and Demonstrations, Industry and Blue Sky Ideas Tracks, ISWC-P and D-Industry-BlueSky 2018

作者： Chen, Haokun Zhang, Xiaowang Feng, Zhiyong School of Computer Science and Technology Tianjin University Tianjin China School of Computer Software Tianjin University Tianjin China Tianjin Key Laboratory of Cognitive Computing and Application Tianjin China

In this paper, we present an ontology-based approach to generate workflows which are core in adaptive data processing by taking advantage of ontologies in characterizing implicit relations among tasks of data processing. Moreover, compared with manual configurations of current approaches, ontology reasoning can automatically infer preference orders of tasks and overcome the limitation of manual configurations. Experimental results show that our proposal is effective and efficient. © 2018 CEUR-WS. All rights reserved.

关键词： Ontology

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：