检索结果-内蒙古大学图书馆

Artificial Intelligence and computer engineering (ICAICE), International Conference on

作者： Yantao Bai Wanwei Liu Xinjun Mao Zhen Liang Key Laboratory of Software Engineering for Complex Systems National University of Defense Technology Changsha China College of Computer Science National University of Defense Technology Changsha China State Key Laboratory of High Performance Computing Institute for Quantum Information &#x0026 National University of Defense Technology Changsha China

ISBN: (纸本)9781665421874

With the development of computer technology, statistics-based machine learning method has made great break-throughs, and also improved the development of artificial intelligence. Nevertheless, as a very influential model, neural networks are still treated as “black boxes”. The results of neural networks are extremely sensitive to the training samples, which lead to great challenges to the controllability of the algorithm. With the wide application of machine learning, demand for interpretability and controllability of neural networks algorithms is increasing. As a result, various scholars have tried to explain and verify neural networks algorithms based on formal methods in recent years. In this paper, a method (called MNNTP) is presented to model the training process of neural networks by using a Markov decision process (MDP). Through MNNTP, the neural networks are abstracted into the form of MDP, which makes notable contributions for verifying some mathematical properties of the neural networks.

关键词： Training Machine learning algorithms Computational modeling Neural networks Machine learning Markov processes Controllability

来源：评论

学校读者我要写书评

暂无评论

Triple-level model inferred collaborative network architecture for video deraining

arXiv

引用

arXiv 2021年

作者： Mu, Pan Liu, Zhu Liu, Yaohua Liu, Risheng Fan, Xin The College of Computer Science and Technology Zhejiang University of Technology Hangzhou China The School of Mathematical Sciences Dalian University of Technology Dalian China The School of Software Dalian University of Technology Dalian116024 China The DUT-RU International School of Information Science & Engineering Dalian University of Technology Dalian116024 China The Pazhou Lab Guangzhou China

Video deraining is an important issue for outdoor vision systems and has been investigated extensively. However, designing optimal architectures by the aggregating model formation and data distribution is a challenging task for video deraining. In this paper, we develop a model-guided triple-level optimization framework to deduce network architecture with cooperating optimization and auto-searching mechanism, named Triple-level Model Inferred Cooperating Searching (TMICS), for dealing with various video rain circumstances. In particular, to mitigate the problem that existing methods cannot cover various rain streaks distribution, we first design a hyper-parameter optimization model about task variable and hyper-parameter. Based on the proposed optimization model, we design a collaborative structure for video deraining. This structure includes Dominant Network Architecture (DNA) and Companionate Network Architecture (CNA) that is cooperated by introducing an Attention-based Averaging Scheme (AAS). To better explore inter-frame information from videos, we introduce a macroscopic structure searching scheme that searches from Optical Flow Module (OFM) and Temporal Grouping Module (TGM) to help restore latent frame. In addition, we apply the differentiable neural architecture searching from a compact candidate set of task-specific operations to discover desirable rain streaks removal architectures automatically. Extensive experiments on various datasets demonstrate that our model shows significant improvements in fidelity and temporal consistency over the state-of-the-art works. Source code is available at https://***/vis-optgroup/TMICS. © 2021, CC BY.

关键词： Network architecture

来源：评论

学校读者我要写书评

暂无评论

MFPN-6D : Real-time One-stage Pose Estimation of Objects on RGB Images

MFPN-6D : Real-time One-stage Pose Estimation of Objects on ...

引用

IEEE International Conference on Robotics and Automation (ICRA)

作者： Penglei Liu Qieshi Zhang Jin Zhang Fei Wang Jun Cheng CAS Key Laboratory of Human-Machine Intelligence-Synergy Systems Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen China Shenzhen College of Advanced Technology University of Chinese Academy of Sciences Beijing China The Chinese University of Hong Kong Hong Kong China The College of Computer Science and Software Engineering Shenzhen University China

6D pose estimation of objects is an important part of robot grasping. The latest research trend on 6D pose estimation is to train a deep neural network to directly predict the 2D projection position of the 3D key points from the image, establish the corresponding relationship, and finally use Pespective-n-Point (PnP) algorithm performs pose estimation. The current challenge of pose estimation is that when the object texture-less, occluded and scene clutter, the detection accuracy will be reduced, and most of the existing algorithm models are large and cannot take the real-time requirements. In this paper, we introduce a Multi-directional Feature Pyramid Network, MFPN, which can efficiently integrate and utilize features. We combined the Cross Stage Partial Network (CSPNet) with MFPN to design a new network for 6D pose estimation, MFPN-6D. At the same time, we propose a new confidence calculation method for object pose estimation, which can fully consider spatial information and plane information. At last, we tested our method on the LINEMOD and Occluded-LINEMOD datasets. The experimental results demonstrate that our algorithm is robust to textureless materials and occlusion, while running more efficiently compared to other methods.

关键词： Deep learning Three-dimensional displays Conferences Pose estimation Grasping Prediction algorithms Market research

来源：评论

学校读者我要写书评

暂无评论

TPFIANet: Three path feature progressive interactive attention learning network for medical image segmentation

引用

Knowledge-Based Systems 2025年 323卷

作者： Yawu Zhao Shudong Wang Yande Ren Jiehuan Wang Shaoqiang Wang Sibo Qiao Tiyao Liu Shanchen Pang Shandong Key Laboratory of Intelligent Oil & Gas Industrial Software Qingdao Institute of Software College of Computer Science and Technology China University of Petroleum Qingdao Shandong China Department of Medical Imaging Center Affiliated Hospital of Qingdao University Qingdao Shandong China Department of Medical Imaging Affiliated Hospital of Jining Medical University Jining Shandong China School of Information and Control Engineering Qingdao University of Technology Qingdao Shandong China School of Software Tiangong University Tianjin China

In the field of medical image segmentation, medical image segmentation methods with high accuracy and generalization capabilities are crucial for driving personalized healthcare solutions and enhancing patient experience. To address the inherent challenges of difficulty in acquiring complex structural features and weak generalization ability in medical image segmentation, we propose a three-path spatial learning (TPSL) module, which takes advantage of convolutional neural networks (CNNs) to skillfully learn specific details at different scales to acquire better feature representations, and then, we propose a progressive feature cross-fusion (PFCF) module, which aims to dynamically learn feature information under different levels in the encoder, which enhances the feature representation. Finally, we propose the feature interactive attention fusion (FIAF) module, which mixes features from different scales and channels, adaptively learns feature weights and enhances the connection between channel and local spatial information. Based on the above modules, we design a three path feature progressive interactive attention learning network (TPFIANet), which captures image features at different levels and scales. We demonstrate the segmentation performance of TPFIANet on three public datasets and three self-organized datasets. Notably, our TPFIANet also demonstrates amiable segmentation performance on datasets of different modalities while maintaining a lighter architecture.

关键词： Feature interactive attention fusion Medical image segmentation Progressive feature cross-fusion Three-path spatial learning

来源：评论

学校读者我要写书评

暂无评论

Comprehensive Analysis of Lightweight Cryptographic Algorithms for Battery-Limited Internet of Things Devices

引用

International Journal of Distributed Sensor Networks 2025年第1期2025卷

作者： Nahom Gebeyehu Zinabu Yihenew Wondie Marye Kula Kekeba Tune Samuel Asferaw Demilew Department of Software Engineering College of Engineering Addis Ababa Science and Technology University Addis Ababa Ethiopia aastu.edu.et Associate Professor in School of Electrical and Computer Engineering (SECE) Addis Ababa Institute of Technology (AAiT) Addis Ababa University (AAU) Addis Ababa Ethiopia aau.edu.et Department of Information Technology College of Computing Debre Berhan University Debre Berhan Ethiopia dbu.edu.et

The growth of billions of devices functioning in resource-constrained situations in the Internet of Things (IoT) era poses serious security issues. As a result of their limited processing power efficiency, memory, and battery life, traditional cryptographic algorithms like elliptic curve cryptography (ECC), Rivest–Shamir–Adleman (RSA), and Advanced Encryption Standard (AES) are sometimes too resource-intensive for these devices. Lightweight cryptography has become a crucial field as a remedy, created especially to maintain security while maximizing energy efficiency and reducing resource usage. With an emphasis on contrasting well-known lightweight block ciphers and pointing out areas for future research, this paper examines the most recent developments in lightweight cryptographic methods. In order to synthesize insights into present issues and future goals, it explores cutting-edge methods that improve performance, security, and energy efficiency. Among the main trends that were covered were the trade-offs between resource limitations and security strength, hardware–software co-optimization, block and stream cipher optimization, and hybrid encryption techniques. The study’s conclusion emphasizes how urgently more research is needed to create cryptographic solutions that can sufficiently safeguard IoT devices.

关键词： battery-limited devices block cipher algorithm embedded system security energy-efficient cryptography IoT security lightweight cryptography low-power cryptographic algorithms and performance analysis

来源：评论

学校读者我要写书评

暂无评论

Edge Caching Replacement Optimization for D2D Wireless Networks via Weighted Distributed DQN

Edge Caching Replacement Optimization for D2D Wireless Netwo...

引用

IEEE Conference on Wireless Communications and Networking

作者： Ruibin Li Yiwei Zhao Chenyang Wang Xiaofei Wang Victor C. M. Leung Xiuhua Li Tarik Taleb College of Intelligence and Computing Tianjin University Tianjin China College of Computer Science and Software Engineering Shenzhen University Shenzhen China College of Intelligence and Computing Chongqing University Chongqing China School of Electrical Engineering Aalto University Helsinki Finland

ISBN: (数字)9781728131061

ISBN: (纸本)9781728131078

Duplicated download has been a big problem that affects the users' quality of service/experience (QoS/QoE) of current mobile networks. Edge caching and Device-to-Device communication are two promising technologies to release the pressure of repeated traffic downloading from the cloud. There are many researches about the edge caching policy. However, these researches have some limitations in the real scenarios. Traditional methods are lacking the self-adaptive ability in the dynamic environment and privacy issues will occur in centralized learning methods. In this paper, based on the virtue of Deep Q-Network (DQN), we propose a weighted distributed DQN model (WDDQN) to solve the cache replacement problem. Our model enables collaboratively to learn a shared predictive model. Trace-driven simulation results show that our proposed model outperforms some classical and state-of-the-art schemes.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Biologically Inspired Feature Enhancement Framework for Zero-Shot Learning

A Biologically Inspired Feature Enhancement Framework for Ze...

引用

IEEE International Conference on Cyber Security and Cloud Computing (CSCloud)

作者： Zhongwu Xie Weipeng Cao Xizhao Wang Zhong Ming Jingjing Zhang Jiyong Zhang College of Computer Science and Software Engineering Shenzhen University Shenzhen China College of Command and Control Engineering Army Engineering University of People’s Liberation Army of China Nanjing China School of Automation Hangzhou Dianzi University Hangzhou China

ISBN: (数字)9781728165509

ISBN: (纸本)9781728165516

Most of the Zero-Shot Learning (ZSL) algorithms currently use pre-trained models as their feature extractors, which are usually trained on ImageNet by using deep neural networks. The richness of the feature information embedded in the pre-trained models can help the ZSL model extract more useful features from its limited training data. However, sometimes the difference between the training data of the current ZSL task and the ImageNet is too large, which may cause the use of pre-trained models has no obvious help or even negative impact on the model performance. To solve this problem, this paper proposes a biologically inspired feature enhancement framework for ZSL. Specifically, we design a dual-channel learning framework that uses auxiliary data sets to enhance the feature extractor of the ZSL model and propose a novel method to guide the selection of the auxiliary data sets based on the knowledge of biological taxonomy. Extensive experimental results show that our proposed method can effectively improve the generalization ability of the ZSL model and achieve state-of-the-art results on three benchmark ZSL tasks. We also explained the experimental phenomena through the way of feature visualization.

关键词： Zero-shot learning feature enhancement feature transfer biological taxonomy ancillary data Respiratory Inspiration extraction device Learning Frameworks Training data Refractive surgery enhancement

来源：评论

学校读者我要写书评

暂无评论

A biologically inspired feature enhancement framework for zero-shot learning

arXiv

引用

arXiv 2020年

作者： Xie, Zhongwu Cao, Weipeng Wang, Xizhao Ming, Zhong Zhang, Jingjing Zhang, Jiyong College of Computer Science and Software Engineering Shenzhen University Shenzhen China College of Command and Control Engineering Army Engineering University of People’s Liberation Army of China Nanjing China School of Automation Hangzhou Dianzi University Hangzhou China

Most of the Zero-Shot Learning (ZSL) algorithms currently use pre-trained models as their feature extractors, which are usually trained on the ImageNet data set by using deep neural networks. The richness of the feature information embedded in the pre-trained models can help the ZSL model extract more useful features from its limited training samples. However, sometimes the difference between the training data set of the current ZSL task and the ImageNet data set is too large, which may lead to the use of pre-trained models has no obvious help or even negative impact on the performance of the ZSL model. To solve this problem, this paper proposes a biologically inspired feature enhancement framework for ZSL. Specifically, we design a dual-channel learning framework that uses auxiliary data sets to enhance the feature extractor of the ZSL model and propose a novel method to guide the selection of the auxiliary data sets based on the knowledge of biological taxonomy. Extensive experimental results show that our proposed method can effectively improve the generalization ability of the ZSL model and achieve state-of-the-art results on three benchmark ZSL tasks. We also explained the experimental phenomena through the way of feature visualization. Copyright © 2020, The Authors. All rights reserved.

关键词： Taxonomies

来源：评论

学校读者我要写书评

暂无评论

A Four-Feature Keyword Extraction Algorithm Based on Word Length Priority Ratio

A Four-Feature Keyword Extraction Algorithm Based on Word Le...

引用

IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom)

作者： Hui Kang Lingfeng Lu Hang Su College of Computer Science and Technology Jilin University Changchun College of Software Engineering Jilin university Changchun Tool product platform department AVIC China Aero-Poly technology Establishment Beijing

ISBN: (数字)9781665403924

ISBN: (纸本)9781665403931

With the rapid development of Internet technology and the advent of the information age, it has become a research hotspot to obtain key information from numerous data. Due to the diversity and irregularity of network data, it is difficult for people to find the literature they want, especially knowledge scholars working in the frontier field of science and technology, who have higher requirements on the accuracy and efficiency of literature keyword extraction than ordinary people. The feature values selected by the current keyword extraction algorithm are usually limited to word frequency and word length, which is incomplete and affects the accuracy of the algorithm. Given this phenomenon, this paper, by comparing with TF-IDF and KEA algorithm, define the concept of word length priority ratio, and applies this concept to the calculation of word length-weight, proposes a four-feature keyword extraction algorithm (WPR-TOC algorithm) based on word frequency, word length, word position and the degree of association between words. Through experiments, compared with the KEA algorithm, KEA++ algorithm, and four features extraction algorithm, the precision of the WPR-TOC algorithm is improved by 40%, 30%, and 10% respectively, and the recall rate is also increased by 40%, 30%, and 10% respectively.

关键词： Knowledge engineering Privacy Feature extraction Information age Eigenvalues and eigenfunctions Internet Security

来源：评论

学校读者我要写书评

暂无评论

AI Enhanced Automatic Response System for Resisting Network Threats 4th

AI Enhanced Automatic Response System for Resisting Network ...

引用

4th International Conference on Smart Computing and Communications, SmartCom 2019

作者： Xia, Song Qiu, Meikang Liu, Meiqin Zhong, Ming Zhao, Hui College of Electronic and Information Wuhan University WuhanHubei China College of Computer Science Shenzhen University ShenzhenGuangdong China Department of Computer Science Harrisburg University of Science and Technology Harrisburg United States College of Electrical Engineering Zhejiang University Zhejiang China School of Software Henan University Kaifeng China

ISBN: (纸本)9783030341381

Network threats are malicious attacks that endanger network security. With terabits of information stored in the network and much of this information being confidential, cyber security turns to be very important. Most network protection mechanisms are based on firewall and Intrusion Detection System (IDS). However, with the diversification of cyber-attacks, traditional defense mechanisms cannot fully guarantee the security of the network. In this paper, we propose an automatic network threat response system based on machine learning and deep learning. It comprises three sub-modules: threat detection module, threat identification module and threat mitigation module. The experimental results show that the proposed system can handle 22 types of network threats in the KDD99 dataset and the rate of successful response is over 97%, which is much better than the traditional ways. © 2019, Springer Nature Switzerland AG.

关键词： Cybersecurity

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：