检索结果-内蒙古大学图书馆

3rd International Conference on Pattern Recognition and Machine Learning, PRML 2022

作者： Huang, Zhongming Wang, Yuxiang Hu, Haolan Liu, Xun Liu, Tongzhen Zhang, Zhanxu Tiangong University School of Electronic and Information Engineering Tianjin China Tiangong University School of Computer Science and Software Tianjin China

ISBN: (数字)9781665499507

ISBN: (纸本)9781665499507

Fire detection technology has been researched and developed for decades. However, in videos and complex scenes, it still lacks fast recognition of fire's existence. The traditional model of fire recognition still needs a large number of samples and time-consuming machine learning progress. Meanwhile, the uncertain shape of fire leads to the reduction of accuracy using CNN. Based on the above problems, we establish a novel method based on I-Vector. We use an adapted I-Vector algorithm to extract the time sequence feature vector on the fire and its surroundings and train a G-PLDA classifier to recognize the dynamic occurrence of fire more quickly and accurately. This model requires fewer samples and a shorter learning time while obtaining an accuracy similar to the traditional fire recognition models, which provides a new effective solution for rapid analysis of whether there is a fire in the video scene. In addition, the algorithm has a sound universality and is easy to deploy in the application fields of video fire supervision, UAV fire inspection, and other related fields. © 2022 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Energy Efficiency Optimization for UAV-Assisted Cellular Networks: A Periodic Clustering-Based MATD3 Approach

Energy Efficiency Optimization for UAV-Assisted Cellular Net...

引用

2024 IEEE Global Communications Conference, GLOBECOM 2024

作者： Liu, Fuhao Chen, Haoqiang Miao, Jiansong Zhang, Tao Zhang, Chuan Kang, Jiawen Niyato, Dusit Beijing University of Posts and Telecommunications School of Information and Communication Engineering Beijing China Beijing Smartchip Microelectronics Technology Company Limited Beijing China Beijing Jiaotong University School of Software Engineering Beijing China Beijing Institute of Technology School of Cyberspace Science and Technology Beijing China Guangdong University of Technology School of Automation Guangzhou China Nanyang Technological University School of Computer Science and Engineering Singapore Singapore

ISBN: (纸本)9798350351255

With the advancement of unmanned aerial vehicles (UAVs) technology, UAV-assisted cellular networks (UACNs) have emerged as a new communication paradigm aimed at enhancing the coverage and capacity of ground networks. Unfortunately, the limited energy capacity of UAVs significantly restricts their operational duration, so optimizing energy efficiency is of importance. However, existing optimization schemes often overlook the impact of ground user mobility on user association, lacking ability to achieve optimal energy efficiency. In this paper, the K-Means method is applied to optimize user association by periodically clustering users. Additionally, given the dynamic nature of the wireless channels, we utilize the Multi-Agent Twin Delayed Deep Deterministic Policy Gradient (MATD3) approach to jointly optimize 3D trajectory and power allocation. The objective is to maximize the sum energy efficiency while meeting the constraints included maximum power, minimum achievable data rate and spatial limitation. Simulation results demonstrate the effectiveness of the proposed algorithm compared with other benchmark algorithms. © 2024 IEEE.

关键词： Mobile telecommunication systems

来源：评论

学校读者我要写书评

暂无评论

A Video Surveillance System Based on Single-Mode Fiber for Energy and Signal Co-Transmission

A Video Surveillance System Based on Single-Mode Fiber for E...

引用

2024 China Automation Congress, CAC 2024

作者： Su, Yuanle Gu, Xueliang Zhang, Zhiguo Wang, Xu Wu, Haoyu Wei, Jialin Song, Xiangning State Key Lab of Information Photonics and Optical Communications Beijing University of Posts and Telecommunications Beijing China School of Electronic Engineering Beijing University of Posts and Telecommunications Beijing China School of Computer Science National Pilot Software Engineering School Beijing University of Posts and Telecommunications Beijing China

ISBN: (纸本)9798350368604

This paper introduces an innovative video surveillance system that utilizes single-mode fiber for the simultaneous transmission of energy and information. Traditional surveillance systems relying on electrical cables pose safety risks and operational challenges, particularly in hazardous and remote environments. The proposed system leverages power-over-fiber (PoF) technology, using a single-mode fiber to co-transmit high-power laser energy and video signals. This approach overcomes the limitations of multimode fibers, such as signal loss and modal dispersion, and enables long-distance transmission up to 10 kilometers. At the remote end, the energy is converted into electrical power by a photovoltaic converter, ensuring the operation of surveillance equipment. The system also incorporates low-power microcontroller (MCU) and a lossless compression algorithm to enhance video transmission efficiency, achieving a frame rate of 16-17 frames per second at a resolution of 320x240. Experimental results demonstrate the system's effectiveness in delivering stable energy supply and high-quality video transmission, making it suitable for use in special environments with stringent safety and operational requirements. © 2024 IEEE.

关键词： Single mode fibers

来源：评论

学校读者我要写书评

暂无评论

Source-Free Unsupervised Domain Adaptation with Sample Transport Learning

引用

Journal of computer science & Technology 2021年第3期36卷 606-616页

作者： Qing Tian Chuang Ma Feng-Yuan Zhang Shun Peng Hui Xue School of Computer and Software Nanjing University of Information Science and TechnologyNanjing 210044China Engineering Research Center of Digital Forensics Ministry of EducationNanjing University of Information Science and TechnologyNanjing 210044China School of Computer Science and Engineering Southeast UniversityNanjing 211189China

Unsupervised domain adaptation(UDA)has achieved great success in handling cross-domain machine learning *** typically benefits the model training of unlabeled target domain by leveraging knowledge from labeled source *** this purpose,the minimization of the marginal distribution divergence and conditional distribution divergence between the source and the target domain is widely adopted in existing ***,for the sake of privacy preservation,the source domain is usually not provided with training data but trained predictor(e.g.,classifier).This incurs the above studies infeasible because the marginal and conditional distributions of the source domain are *** this end,this article proposes a source-free UDA which jointly models domain adaptation and sample transport learning,namely Sample Transport Domain Adaptation(STDA).Specifically,STDA constructs the pseudo source domain according to the aggregated decision boundaries of multiple source classifiers made on the target ***,it refines the pseudo source domain by augmenting it through transporting those target samples with high confidence,and consequently generates labels for the target *** train the STDA model by performing domain adaptation with sample transport between the above steps in alternating manner,and eventually achieve knowledge adaptation to the target domain and attain confident labels for ***,evaluation results have validated effectiveness and superiority of the proposed method.

关键词： unsupervised domain adaptation domain shift sample transport pseudo source domain

来源：评论

学校读者我要写书评

暂无评论

Intent-aware Prompt Learning for Medical Question Summarization

Intent-aware Prompt Learning for Medical Question Summarizat...

引用

2022 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2022

作者： Zhang, Leilei Liu, Junfei Peking University School of Software & Microelectronics Beijing China Peking University School of Electronics Engineering and Computer Science Beijing China

ISBN: (纸本)9781665468190

Consumer health question summarization aims to generate concise questions expressing the minimum information needed to find the correct answer to the original user queries, which can significantly improve the success rate of retrieving relevant answers in automated medical question answering systems. Recently, prompt learning has emerged as a new paradigm of leveraging pre-trained language models (PLMs) with promising results in downstream tasks. Existing prompt-based methods assume that all samples in a task share the same prompt. However, the health questions raised by consumers in search engines focus on different types of medical entities and cover different purposes. It is not sufficient to simply use the same prompt regardless of consumer intent awareness, which likely leads to meaningless or substandard quality summaries. In this paper, we propose an intent-aware prompt learning (InPL) method, which consists of a question intent recognition module, a dynamic prompt module and a medical summarization generation module to model the medical summarization task in a unified transformer. Our InPL architecture generates high-quality summaries capturing more consumer health intent and encourages machines to generate more factually correct summaries. Extensive experiments on the popular medical summarization datasets demonstrate the superiority of our approach. © 2022 IEEE.

关键词： Search engines

来源：评论

学校读者我要写书评

暂无评论

Fusion of speech and handwritten signatures biometrics for person identification

引用

International Journal of Speech Technology 2023年第4期26卷 833-850页

作者： Abushariah, Ahmad A. M. Abushariah, Mohammad A. M. Gunawan, Teddy Surya Chebil, J. Alqudah, Assal A. M. Ting, Hua-Nong Mustafa, Mumtaz Begum Peer Faculty of Engineering University of Malaya Kuala Lumpur50603 Malaysia Department of Computer Information Systems King Abdullah II School of Information Technology The University of Jordan Amman Jordan ECE Department Faculty of Engineering International Islamic University Malaysia Kuala Lumpur53100 Malaysia Department of Computer Science Faculty of Science and Information Technology Al-Zaytoonah University of Jordan Amman Jordan Department of Biomedical Engineering Faculty of Engineering University of Malaya Kuala Lumpur50603 Malaysia Department of Software Engineering Faculty of Computer Science and Information Technology University of Malaya Kuala Lumpur50603 Malaysia

Automatic person identification (API) using human biometrics is essential and highly demanded compared to traditional API methods, where a person is automatically identified using his/her distinct characteristics including speech, fingerprint, iris, handwritten signatures, and others. The fusion of more than one human biometric produces bimodal and multimodal API systems that normally outperform single modality systems. This paper presents our work towards fusing speech and handwritten signatures for developing a bimodal API system, where fusion was conducted at the decision level due to the differences in the type and format of the features extracted. A data set is created that contains recordings of usernames and handwritten signatures of 100 persons (50 males and 50 females), where each person recorded his/her username 30 times and provided his/her handwritten signature 30 times. Consequently, a total of 3000 utterances and 3000 handwritten signatures were collected. The speech API used Mel-Frequency Cepstral Coefficients (MFCC) technique for features extraction and Vector Quantization (VQ) for features training and classification. On the other hand, the handwritten signatures API used global features for reflecting the structure of the hand signature image such as image area, pure height, pure width and signature height and the Multi-Layer Perceptron (MLP) architecture of Artificial Neural Network for features training and classification. Once the best matches for both the speech and the handwritten signatures API are produced, the fusion process takes place at decision level. It computes the difference between the two best matches for each modality and selects the modality of the maximum difference. Based on our experimental results, the bimodal API obtained an average recognition rate of 96.40%, whereas the speech API and the handwritten signatures API obtained average recognition rates of 92.60% and 75.20%, respectively. Therefore, the bimodal API system is a

关键词： Biometrics

来源：评论

学校读者我要写书评

暂无评论

An Efficient Classification Techniques for Brain Tumor Using Features Extraction and Statistic Methods, with Machine Learning Algorithms

An Efficient Classification Techniques for Brain Tumor Using...

引用

作者： Badshah, Shah Hussain Farhatullah khan, Gul Zaman Hassan, Muhammad Abul Junaid, Hazrat Sohail, Muhammad Mahbob, Muhammad Awais Ahamad, Izaz Ullah, Nadeem Department of Computing and Technology Abasyn University Peshawar Pakistan School of Automation China University of Geosciences Wuhan430074 China Department of Computer Software Engineering University of Engineering and Technology Mardan Pakistan Department of Computer Science and Information Technology University of Malakand Dir Lower Pakistan Department of Computer Science Bahria University Islamabad Pakistan Department of Computer Science University of Engineering and Technology Taxila Pakistan Department of Information Engineering and Computer Science University of Trento Trento38122 Italy

Today, brain tumor is a very dangerous disease that can even cause death. There are many ways to classify Brain MRI images of tumors. Various aspects of current research have limitations;some methods are accurate but take a long time to compute while some algorithms are fast, but with low accuracy. Consequently, a lot of effort is needed in this area. In this paper, our major goal is to reduce computation time and improve the classification accuracy for brain MRI images. This research proposes four phases: preprocessing, feature extraction, feature selection, and classification. During pre-processing, the median filter was used to reduce noise in the images, and the grayscale image size was reduced to further limit the image for efficient use. After that the Gray images are then given as inputs to the Feature extraction phase to select a limited feature from the image namely, texture features, shape features, smoothing features. The images features obtained in the feature’s extraction phase are still large, and we cannot simply feed them to machine learning models for processing due to computational constraints. So, we need to analyze it and pick out some interesting elements from the images. The statistical features namely, variances, mean, correlation are used in the feature selection phase to select more significant features to reduce the size of the features and reduce computational time. These features are combined and labeled in a file to train Machine Learning (ML) algorithms. We used two types of machine learning algorithms in the classification phase: classifier and evolutionary algorithm. The selected classifier algorithms results are combined in a file and we named it S-K-E-L Classifier and it gives us acceptable results in terms of accuracy and time. Similarly, an evolutionary algorithm named P-S-D-G Classifier also gives us acceptable results in terms of accuracy and time. Lastly, we combine both S-K-E-L Classifier and P-S-D-G Classifier in a file with

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

Evaluating and transforming you and your colleagues' assessments in an age of generative AI 52

Evaluating and transforming you and your colleagues' assessm...

引用

52nd Annual Conference of the European Society for engineering: Educating Responsible Engineers, SEFI 2024

作者： Neal, P.R. Daniel, S. Hassan, G.M. Grundy, S. Haque, R. Nikolic, S. Belkina, M. Lyden, S. Sandison, C. School of Chemical Engineering University of New South Wales Sydney Australia School of Professional Practice and Leadership University of Technology Sydney Sydney Australia School of Computer Science and Software Engineering University of Western Australia Perth Australia School of Science Technology and Engineering University of the Sunshine Coast Sippy Downs Australia Faculty of Engineering and Information Sciences University of Wollongong Wollongong Australia College Western Sydney University Sydney Australia School of Engineering University of Tasmania Hobart Australia

ISBN: (纸本)9782873520274

ChatGPT and other Generative AI tools (GenAI) have generated much commotion and confusion within academic circles. Many academics still need to understand the risk such tools pose on current assessment practices and how students can use them for improved grades/outcomes. Unfortunately, most studies are very generic, do not go into much detail, and are outside of the engineering education context. In addition, GenAI provides many opportunities to construct novel, authentic and/or personalised learning experiences for students. The workshop's facilitators have completed two comprehensive studies evaluating ChatGPT's impact on engineering education assessment (Nikolic et al. 2023;Nikolic et al. 2024). While strengths, weaknesses and opportunities were outlined, it is only the starting point for much-needed conversations. This workshop introduced participants to various GenAI tools in the context of a variety of engineering assessment types and engaged participants with a range of experience levels. We guided and engaged participants in exploring GenAI capabilities, in applying a framework for evaluating security and opportunities in assessment and through this demonstrated a model for professional development that they can adapt to their own institutions. © 2024 SEFI 2024 - 52nd Annual Conference of the European Society for engineering, Proceedings: Educating Responsible Engineers. All rights reserved.

关键词： Students

来源：评论

学校读者我要写书评

暂无评论

Curriculum Contrastive Learning for COVID-19 FAQ Retrieval

Curriculum Contrastive Learning for COVID-19 FAQ Retrieval

引用

2022 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2022

作者： Zhang, Leilei Liu, Junfei Peking University School of Software & Microelectronics Beijing China Peking University School of Electronics Engineering and Computer Science Beijing China

ISBN: (纸本)9781665468190

Medical Frequently Asked Question (FAQ) retrieval aims to find the most relevant question-answer pairs for a given user query, which is of great significance for enhancing people medical health awareness and knowledge. However, due to medical data privacy and labor-intensive labeling, there is a lack of large-scale question-matching training datasets. Previous methods directly use the collected question-answer pairs on search engines to train retrieval models, which achieved poor performance. Inspired by recent advances in contrastive learning, we propose a novel contrastive curriculum learning framework for modeling user medical queries. First, we design different data augmentation methods to generate positive samples and different types of negative samples. Second, we propose a curriculum learning strategy that associates difficulty levels with negative samples. Through a contrastive learning process from easy to hard, our method achieves excellent results on two medical datasets. © 2022 IEEE.

关键词： Curricula

来源：评论

学校读者我要写书评

暂无评论

A Nudity Detection Algorithm for Web-based Online Networking Platform 10

A Nudity Detection Algorithm for Web-based Online Networking...

引用

17th INDIACom;2023 10th International Conference on Computing for Sustainable Global Development, INDIACom 2023

作者： Dewan, Ritu Polishetty, Rohith Jagadam, Naveen Ravulakollu, Kiran Kumar Goyal, Mayank Kumar Sharan, Bhagwati Galgotias College of Engineering and Technology Department of Computer Science and Engineering Greater Noida India Sr. Software Engineer Migo India Limited Hyderabad India Ceo N2force Technologies Inc. NapervilleIL60563 United States School of Technology Woxsen University Hyderabad India School of Engineering and Technology Sharda University Department of Computer Science and Engineering Greater Noida India School of Engineering and Sciences Srm University Department of Computer Science and Engineering AP Amaravati India

ISBN: (纸本)9789380544472

Social media and online platforms have become an integral part of the day-to-day life of the people and the organizations, which shows a lot of diversity. Many online platforms allow users to share media extensively, like images, videos, and audio but not all the media content shared on the internet is safe for work or should be restricted from particular platforms. Images with nudity are of big concern when it comes to online media sharing and social platforms must take serious measures to restrict users to share images with nudity. The rate of growth of nudity content over the internet or social media is very high. It affects the mental and physical health of individuals. In this paper, a novel method for nudity detection in images is presented. We aim to develop a Machine Learning algorithm for nudity detection in a community networking platform and how it should be integrated with such a platform so that posted images can be thoroughly checked for nudity and if detected how it should be dealt with. © 2023 Bharati Vidyapeeth, New Delhi.

关键词： Social networking (online)

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：