检索结果-内蒙古大学图书馆

SSRN 2023年

作者： Pinasthika, Krisna Laksono, Blessius Sheldo Putra Irsal, Riyandi Banovbi Putera Shabiyya, Syifa' Hukma Yudistira, Novanto Faculty of Computer Science Informatics Engineering Department Brawijaya University Malang65145 Indonesia

Advancements in computer vision research have put transformer architecture as the state-of-the-art in computer vision tasks. One of the known drawbacks of the transformer architecture is the high number of parameters, this can lead to a more complex and inefficient algorithm. This paper aims to reduce the number of parameters and in turn, made the transformer more efficient. We present Sparse Transformer (SparTa) Block, a modified transformer block with an addition of a sparse token converter that reduces the number of tokens used. We use the SparTa Block inside the Swin-T ar-chitecture (SparseSwin) to leverage Swin’s capability to downsample its input and reduce the number of initial tokens to be calculated. The proposed SparseSwin model outperforms other state-of-the-art models in image classification with an accuracy of 86.96%, 97.43%, and 85.35% on the ImageNet100, CIFAR10, and CIFAR100 datasets respectively. Despite its fewer parameters, the result highlights the potential of a transformer architecture using a sparse token converter with a limited number of tokens to optimize the use of the transformer and improve its performance. © 2023, The Authors. All rights reserved.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

Human-Robot Interaction on Elderly Companion Robot Development Using Dual Intent Entity Transformer

Human-Robot Interaction on Elderly Companion Robot Developme...

引用

International Conference on Information Technology, Information Systems and Electrical engineering (ICITISEE)

作者： Naufal Haidar Rauf Heru Agus Santoso Department of Informatics Engineering Faculty of Computer Science Universitas Dian Nuswantoro Semarang Indonesia

The growing elderly population and the increasing demand for nursing home care have led to a need to improve the quality of life for residents. A popular solution is developing a companion robot to assist residents with various tasks and provide human-like interaction. In this paper, we present the prototype of a companion robot currently website-based, equipped with a Dual Intent Entity Transformer to understand what the user wants. The results of the study show the prototype is capable of classifying user messages according to their appropriate intent with precision, recall, F-score, and accuracy on average 91 %, 91 %, 90%, and 87% respectively using privately owned dataset. Future works such as expanding the robot's insight and tuning the model would improve the service provided by the companion robot.

关键词：

来源：评论

学校读者我要写书评

暂无评论

No-Reference Image Quality Assessment Using Meta-Learning 3rd

No-Reference Image Quality Assessment Using Meta-Learning

引用

The 3rd International Conference on Frontiers in Computing and Systems (COMSYS 2022)

作者： Dey, Ratnadeep Bhattacharjee, Debotosh Kejcar, Ondrej Department of Computer Science and Engineering Jadavpur University Kolkata India Center for Basic and Applied Science Faculty of Informatics and Management University of Hradec Kralove Rokitanskeho 62 Hradec Kralove500 03 Czech Republic

ISBN: (纸本)9789819926794

Deep learning-based no-reference image quality assessment faces problems like dependency on a large amount of experimental data and the generalization ability of the learned model. A deep learning model trained on a specific dataset cannot obtain the desired results for testing on other datasets. Similarly, a deep learning model trained with small experiment data does not provide the best result. This paper addresses these problems of the deep learning model using the meta-learning approach in the field of no-reference Image Quality Assessment. The no-reference image quality assessment is a small sample problem, where the amount of experimental data is very less. Although data augmentation techniques have been used to increase the amount of experimental data, they do not increase the variation of the data. Therefore, traditional deep learning-based techniques are unsuitable for no-reference image quality assessment. Another problem is the lack of generalization ability. A deep learning model trained with image quality datasets containing images distorted synthetically cannot efficiently assess the quality of images distorted naturally. This work proposes a meta-learning model that can be trained with more than one image quality dataset, where one image quality assessment dataset contains synthetic images and the other contains real images. Finally, another image quality assessment dataset has tested the trained model. The test result is better than the state-of-the-art methods, and the results establish the fact that the meta-learning model proposed in this paper tries to resolve the problems of the deep learning model. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Image quality

来源：评论

学校读者我要写书评

暂无评论

Distributed Gradient Clustering: Convergence and the Effect of Initialization 58

Distributed Gradient Clustering: Convergence and the Effect ...

引用

58th Asilomar Conference on Signals, Systems and computers, ACSSC 2024

作者： Armacki, Aleksandar Sharma, Himkant Bajović, Dragana Jakovetić, Dušan Chakraborty, Mrityunjoy Kar, Soummya Carnegie Mellon University PittsburghPA United States Indian Institute of Technology Kharagpur Kharagpur India University of Novi Sad Faculty of Technical Sciences Department of Power Electronics and Computer Engineering Novi Sad Serbia University of Novi Sad Faculty of Sciences Department of Mathematics and Informatics Novi Sad Serbia

ISBN: (纸本)9798350354058

We study the effects of center initialization on the performance of a family of distributed gradient-based clustering algorithms introduced in [1], that work over connected networks of users. In the considered scenario, each user contains a local dataset and communicates only with its immediate neighbours, with the aim of finding a global clustering of the joint data. We perform extensive numerical experiments, evaluating the effects of center initialization on the performance of our family of methods, demonstrating that our methods are more resilient to the effects of initialization, compared to centralized gradient clustering [2]. Next, inspired by the K-means++ initialization [3], we propose a novel distributed center initialization scheme, which is shown to improve the performance of our methods, compared to the baseline random initialization. © 2024 IEEE.

关键词： K-means clustering

来源：评论

学校读者我要写书评

暂无评论

LSTM deep learning model for Alzheimer's disease prediction based on cost-effective time series cognitive scores 5

LSTM deep learning model for Alzheimer's disease prediction ...

引用

5th International IEEE Novel Intelligent and Leading Emerging Sciences Conference, NILES 2023

作者： Saleh, Hager El-Rashidy, Nora Abuhmed, Tamer El-Sappagh, Shaker South Valley University Faculty of Computers and Artificial Intelligence Hurghada Egypt Kafrelsheiksh University Faculty of Artificial Intelligence Machine Learning and Information Retrieval Department Egypt Sungkyunkwan University College of Computing and Informatics Korea Republic of Galala University Faculty of Computer Science and Engineering Suez435611 Egypt

ISBN: (纸本)9798350381030

Alzheimer's disease (AD) is a complex chronic neurodegenerative disease that propagates over time. Deep learning (DL) models can be used to learn time series data to extract deep temporal features and make robust decisions. The fusion of multimodal time series data has been proven to enhance model performance [1]. For instance, cognitive scores (CSs) including clinical dementia rating, and Alzheimer's disease assessment scores have been integrated with other modalities like MRI to predict the future status of AD patients [2]. They enhanced the performance of DL models. These scores can be collected easily and in a cost-effective way in hospitals. No study in the literature has built a deep learning model to predict AD based on the cost-effective and multimodal time series data of CSs. In this study, we propose an LSTM-based DL architecture to predict AD based on multiple time series CSs. The proposed model has been optimized using a Bayesian optimizer to select the best architecture, and it has been compared with multiple classical machine learning models like random forest and others. The proposed LSTM architecture achieved better results than other models and provided stable and robust performance. © 2023 IEEE.

关键词： Data fusion

来源：评论

学校读者我要写书评

暂无评论

A Fused Machine Learning Approach for Intrusion Detection System

引用

computers, Materials & Continua 2023年第2期74卷 2607-2623页

作者： Muhammad Sajid Farooq Sagheer Abbas Atta-ur-Rahman Kiran Sultan Muhammad Adnan Khan Amir Mosavi School of Computer Science National College of Business Administration and EconomicsLahore54000Pakistan Department of Computer Science College of Computer Science and Information TechnologyImam Abdulrahman Bin Faisal UniversityP.O.Box 1982Dammam 31441Saudi Arabia Department of CIT The Applied CollegeKing Abdulaziz UniversityJeddah21589Saudi Arabia Department of Software Gachon UniversitySeongnam13120Korea John von Neumann Faculty of Informatics Obuda UniversityBudapest1034Hungary Institute of Information Engineering Automation and MathematicsSlovak University of Technology in BratislavaBratislava81107Slovakia Faculty of Civil Engineering TU-DresdenDresden01062Germany

The rapid growth in data generation and increased use of computer network devices has amplified the infrastructures of *** interconnectivity of networks has brought various complexities in maintaining network availability,consistency,and *** learning based intrusion detection systems have become essential to monitor network traffic for malicious and illicit *** intrusion detection system controls the flow of network traffic with the help of computer *** deep learning algorithms in intrusion detection systems have played a prominent role in identifying and analyzing intrusions in network *** this purpose,when the network traffic encounters known or unknown intrusions in the network,a machine-learning framework is needed to identify and/or verify network *** Intrusion detection scheme empowered with a fused machine learning technique(IDS-FMLT)is proposed to detect intrusion in a heterogeneous network that consists of different source networks and to protect the network from malicious *** proposed IDS-FMLT system model obtained 95.18%validation accuracy and a 4.82%miss rate in intrusion detection.

关键词： Fused machine learning heterogeneous network intrusion detection

来源：评论

学校读者我要写书评

暂无评论

Proposing a Novel Framework for Prediction of Stock using Machine Learning 11

Proposing a Novel Framework for Prediction of Stock using Ma...

引用

11th International Conference on Cyber and IT Service Management, CITSM 2023

作者： Bist, Ankur Singh Rawat, Bhupesh Kosasi, Sandy Aini, Qurotul Oganda, Fitra Putri Yadila, Ahmad Bayu Graphic Era Hill University Uttarakhand India Stmik Pontianak Faculty of Informatics Engineering West Kalimantan Indonesia University of Raharja Information System Tangerang Indonesia University of Raharja Computer System Tangerang Indonesia

ISBN: (纸本)9798350305968

Stock prices are highly volatile, dynamic, and non-linear, making it very difficult to predict the exact price at any given time. In addition, stock prices are influenced by several factors, such as political conditions, the global economy, unexpected events, company financial performance, and more. Up to this point, various machine learning techniques have been employed for stock prediction;however, none of these techniques can accurately predict stock prices due to the high volatility in stock prices. Machine learning approaches, such as random forest, SVM, KNN, and logistic regression, represent some of the algorithms used for stock prediction. This paper aims to propose a new framework based on machine learning and deep learning for stock prediction. The prediction relies on the company's stock fundamentals, past performance, related stocks in the same sector, and other relevant factors. © 2023 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Performance Analysis of the Edge-Preservation based Normalized Convolution for Medical Image Enhancement and Noise Reduction

Performance Analysis of the Edge-Preservation based Normaliz...

引用

Biomedical Instrumentation and Technology Conference (IBITeC), International

作者： Aris Marjuni Oky Dwi Nurhayati Department of Informatics Engineering Faculty of Computer Science Universitas Dian Nuswantoro Semarang Indonesia

Acquired medical images often contain noise due to various factors that affect image quality. The noisy image indicates the presence of inappropriate information or loss of original information due to changes in the pixel values. For diagnosis purposes, the acquired or noisy medical image needs to be reconstructed to obtain an enhanced image or denoised image for a better decision. One of the techniques to enhance or reduce the noise can be performed by normalized convolution to interpolate the change of pixels in the original image. This paper is presented to investigate the feasibility of the normalized convolution filter to enhance and reduce medical image noise. The study was performed through experiments using several types of medical images, namely Brain MRI, Breast Ultrasound, Chest X-ray, Retinal OCT, Mammogram, and RSNA Bone Age images. Experimental results show that the normalized convolution could produce an enhanced image with sharper contrast. In the image enhancement, the enhanced image obtained by this filter achieves 41.12 dB on the PSNR average. In the medical image denoising, the normalized convolution could produce better visual quality to reduce the Gaussian, Salt & Pepper, Poisson, and Speckle with PSNR averages are 28.43dB, 32.82dB, 32.49dB, and 29.30dB. This method also outperforms basic image denoising filters, such as Imbox, Wiener, and Median Filterings. Hence, this filter could be used as an alternative for medical image preprocessing, such as image enhancement and noise reduction.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Quantifying Object-Oriented System Complexity: Introducing a Powerful Measurement Tool 13

Quantifying Object-Oriented System Complexity: Introducing a...

引用

13th IEEE International Conference on System engineering and Technology, ICSET 2023

作者： Jasser, Muhammed Basheer Zhen, Lee Ming Issa, Bayan Hong, Ling Mee Al-Hadi, Ismail Ahmed Al-Qasem Sunway University Department of Computing and Information Systems Selangor Sunway City Malaysia University of Aleppo Faculty of Informatics Engineering Syria Institute of Computer Science & Digital Innovation UCSI University Kuala Lumpur Malaysia

ISBN: (纸本)9798350340891

A lot of metrics and tools have been devised to measure the complexity of software systems. Through measurement, software professionals can have a better understanding and control of a software's complexity. In this paper, we reviewed complexity metrics and tools for object-oriented systems. The existing tools measure either implemented systems which may be source code or compile binaries, or class diagrams. The tools that measure implemented systems already offer a wide variety of features to satisfy different needs. However, the tools that measure class diagrams are not only poor in amount, but all take in class diagram XMI files, with none of them taking in PlantUML class diagram source code. PlantUML is a popular text-based UML diagramming tool that enables users to define UML diagrams with a domain specific language then renders the diagram for them. Hence, we developed a web application that measures class diagram complexity via PlantUML class diagram source code. It measures five class metrics and seven system metrics visualized as tables and bar charts, with the results exportable to PDF and CSV. Our tool's measurements were also evaluated to be correct when it is within our scope of supported syntax. © 2023 IEEE.

关键词： Object oriented programming

来源：评论

学校读者我要写书评

暂无评论

Deep Learning-based Method for Automatic Fruit Fly Detection and Counting

Deep Learning-based Method for Automatic Fruit Fly Detection...

引用

Computing (ICOCO), IEEE International Conference on

作者： Indah Desri Wahyuni Nabila Husna Shabrina Suputa Department of Computer Engineering Faculty of Engineering and Informatics Universitas Multimedia Nusantara Tangerang Indonesia Department of Plant Protection Faculty of Agriculture Universitas Gadjah Mada Yogyakarta Indonesia

ISBN: (数字)9798331530303

ISBN: (纸本)9798331530310

Fruit flies are a major threat faced by snake fruit farmers. Fruit flies can degrade the quality of snake fruits and reduce the overall yield. Traps stuffed with Methyl Eugenol are commonly placed across snake fruit plantations to attract fruit flies. Farmers then manually counted and reported the trapped fruit flies as part of their pest management strategies. However, manually counting trapped fruit flies is time-consuming, leading to unwillingness among farmers to perform this task. This study aims to develop an automated detection and counting system using deep-learning-based methods to address this issue. The proposed system was built on a modified YOLOv5 algorithm, achieving a precision of 0.944, recall of 0.947, and mAP of 0.946, with an object-counting time of only 50 ms. These results demonstrate that the proposed fruit-fly detection and counting system offers excellent performance with rapid detection times

关键词： YOLO Training Learning systems Deep learning Plantations Pest control Inference algorithms Mobile applications Monitoring Testing

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：