检索结果-内蒙古大学图书馆

2024 International Conference on Advances in Computing, Communication, Electrical, and Smart Systems, iCACCESS 2024

作者： Laila, Shahara Ema, Romana Rahman Galib, Syed Md. Jamil, Abu Rafe Md Uddin, A. F. M. Shahab Jashore University of Science and Technology Dept. of Computer Science and Engineering Jashore Bangladesh

ISBN: (纸本)9798350350289

Oral carcinoma affects millions of individuals globally and is a major public health problem due to its high fatality rate. Treatment effectiveness depends on early detection and improves survival rates. An effective method for automated oral cancer screening is deep learning. The paper's primary goal is to use deep learning techniques to advance the state of oral cancer diagnosis, specifically YOLOv5 and YOLOv7. We aimed to achieve accurate detection by leveraging bounding box annotations for both two distinct types of data: collected clinical data and benchmark data of oral carcinoma. The proposed methodology encompasses data collection, preprocessing, expert-driven bounding box annotation and the use of cutting-edge deep learning models for lesion identification, YOLOv5 and YOLOv7. Bounding box serving as expert annotations are meticulously added to pinpoint cancerous regions within the images. Subsequently, a comprehensive evaluation is conducted, measuring the performance of both models based on precision, recall rate (sensitivity) and mean average precision (mAP). Our findings reveal that YOLOv5 outperforms YOLOv7 in terms of oral cancer detection, demonstrating superior precision and recall rates. In the evaluation of this research, the YOLOv5 model demonstrated strong performance metrics on both clinical and lip and tongue cancer datasets. For the lip and tongue cancer dataset, the YOLOv5 model achieved a precision of 97.2%, a mAP (mean Average Precision) of 97.7% and a recall of 96.3%. © 2024 IEEE.

关键词： Clinical research

来源：评论

学校读者我要写书评

暂无评论

Beyond Copy Paste: Plagiarism Detection using Machine Learning 7

Beyond Copy Paste: Plagiarism Detection using Machine Learni...

引用

7th International Conference on Inventive Computation Technologies, ICICT 2024

作者： Vasuteja, Akula Reddy, Alavalapalli Vasanth Pravin, A. Sathyabama Institute of Science And Technology Dept. of Computer Science And Engineering Chennai India

ISBN: (纸本)9798350359299

Since plagiarism is a major problem in both academic and professional settings, it is crucial to create effective plagiarism detection methods. Accuracy evaluation for classical models such as SVM, Naive Bayes, and Random Forests usually involves cross-validation, where the dataset is split into training and testing folds, and the accuracy for each fold is then computed. These techniques lack the neural network-like loss functions, but they can still be used to analyze model performance by charting accuracy against the number of folds. But in CNNs and RNNs, training entails updating a loss function over epochs through iterative optimization;for classification problems, this is usually categorical cross-entropy. The values of the loss function are monitored during training, and the convergence of the model is shown by graphing them against the number of epochs. Accuracy metrics are calculated in parallel for the training and validation stages, and the model's classification accuracy throughout successive epochs is displayed in graphs. The plagiarism detector provides users with a complete report that highlights plagiarized portions together with their original sources which results in 97.99% of accuracy in order to aid educators, researchers, and other professionals in preventing plagiarism. © 2024 IEEE.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Real Time Monitoring and Configuration Interface for Smart Home Device Management Using Streamlit and Pygame

Real Time Monitoring and Configuration Interface for Smart H...

引用

2024 International Conference on Advances in Data engineering and Intelligent Computing Systems, ADICS 2024

作者： Mani, G. Suma Krishna, G. Venkata Praveena, M.D. Anto Asha, P. Selvan, Mercy Paul Sathyabama Institute Of Science And Technology Dept. Of Computer Science And Engineering Chennai India

ISBN: (纸本)9798350364828

One of the helpful tool for measuring and managing home power consumption is one of the clever and creative tool for tracking the usage of Home Power Management System. The Goal of this system is to lower the power bills, also promote energy efficiency, and provide the users active control over how much energy they use. Smart meters, an intuitive and user-friendly smartphone app, and real-time energy monitoring sensors are some of the system's essential parts and these are the key parts of the system. Installing smart meters makes it possible to accurately track the power usage and provide data to the system. Users can be able to see their energy usage instantly and make smart decisions when real-time energy monitoring equipment is included. Further, the system uses advanced smart technology, analytics and machine learning algorithms to provide clients customized suggestions on how to save energy consumption. Homes may save a lot of money on electricity bills, help the environment, and have more control over how much energy they use by using our Smart Home Power Management System. © 2024 IEEE.

关键词： Smart meters

来源：评论

学校读者我要写书评

暂无评论

Effective Polyp Segmentation in Medical Imaging: A Semi-Supervised Approach with Progressive Supervision 3

Effective Polyp Segmentation in Medical Imaging: A Semi-Supe...

引用

3rd International Conference on Big Data, Information and computer Network, BDICN 2024

作者： Ye, Shunlong Ji, Zexuan Nanjing University of Science and Technology Dept. of Computer Science and Engineering Nanjing China

ISBN: (纸本)9798350363821

Medical imaging plays a pivotal role in diagnosing and monitoring various diseases, and accurate polyp segmentation is essential for early disease detection and treatment planning. However, the scarcity of well-Annotated medical imaging data hampers the development of effective segmentation models. Semi-supervised learning stands out as a critical approach to address this issue. Traditional semi-supervised medical image segmentation models leverage strong-weak perturbation consistency losses to provide supervision signals. While these losses are effective in promoting consistency, they often exhibit a drawback, that is the dominance of strong perturbations during early training stages. This dominance hinders the model's ability to extract meaningful features, ultimately impacting its overall performance. In response to these challenges, we introduce a novel training method applied to polyp segmentation. We employ a progressive approach, gradually introducing strong-weak perturbation supervision signals. This gradual guidance empowers the model to focus on learning relevant features from the beginning of training, mitigating the issue of early dominance and enhancing its performance. Our method is evaluated on two widely-used polyp segmentation datasets and surpasses state-of-The-Art methods, demonstrating substantial improvements in segmentation accuracy. © 2024 IEEE.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Personalized Video Prefetching Mechanism Based on User Viewing History in Edge Computing Environment

引用

Transactions of the Korean Institute of Electrical Engineers 2023年第5期72卷 620-627页

作者： Moon, YangChan Lim, Mingyu Dept. of Smart ICT Convergence Konkuk University Korea Republic of Dept. of Computer Science and Engineering Konkuk University Korea Republic of

The personalized video service collects and analyzes user information to provide differentiated services for each user. Additionally, it is impossible to avoid delays when sending a video from a server to a client. However, if prefetching is performed at the edge node, transmission delay can be reduced, and personalized service can be provided by utilizing user information at each edge node. In this paper, we propose a personalized video service based on edge nodes. The proposed method operates so that the edge node performs personalized prefetching in which the number of prefetching is adjusted by calculating the weights of five selection criteria based on the user's viewing record. We found that the proposed method has the highest prefetching accuracy through implementation and experiments using the simulation of user behavior of selecting and viewing videos. © 2023 Korean Institute of Electrical Engineers. All rights reserved.

关键词： Edge computing

来源：评论

学校读者我要写书评

暂无评论

Real-Time Dashboarding using Big Data Tools 7

Real-Time Dashboarding using Big Data Tools

引用

7th International Conference on Inventive Computation Technologies, ICICT 2024

作者： Sanjay, R. Pulakhandam, Dheeraj Nirmalrani, V. Sathyabama Institute of Science And Technology Dept. Computer Science And Engineering Chennai India

ISBN: (纸本)9798350359299

Real-time dashboarding software empowers organizations to harness the burgeoning IoT data stream, transforming raw sensor readings into actionable insights through a secure and scalable architecture. Data from connected devices seamlessly flows via lightweight protocols like MQTT and is managed by a robust streaming platform like Kafka. Big data tools like Spark then extract meaning from this high-velocity data, preparing it for visualization on dynamic dashboards hosted by cloud platforms. These user-centric dashboards track key metrics, identify trends, and drive informed decision-making, but tackling challenges like scalability, security, and data quality is crucial to unlock the full potential of this transformative technology. © 2024 IEEE.

关键词： Data visualization

来源：评论

学校读者我要写书评

暂无评论

Revolutionizing Road Safety for Advanced Vehicular Accident Prediction 2

Revolutionizing Road Safety for Advanced Vehicular Accident ...

引用

2nd International Conference on Intelligent Data Communication Technologies and Internet of Things, IDCIoT 2024

作者： Deepika, T. Yuvaraj, N. Sri Preethaa, K.R. KPR Institute of Engineering and Technology Dept. of Computer Science and Engineering Coimbatore India

ISBN: (纸本)9798350327533

The detection and tracking of vehicle speed are crucial for ensuring the safety of civilians and preventing accidents. Speed tracking plays a pivotal role in traffic surveillance, where effective management and the well-being of citizens take precedence. The approach leverages computer vision-based models, specifically Multi-scale Detector for object detection, and a fusion of the DeepSORT and Centroid Sort algorithms for vehicle tracking from streaming video. To cater to the requirements of these algorithms, a custom dataset was developed within a controlled setting. Dataset augmentation was employed to prevent overfitting, thereby enhancing the dataset's diversity and the model's ability to generalize. The method demonstrates the capacity to accurately detect and track vehicles, maintaining performance even in high-traffic conditions. Cross-validation at every epoch during the training phase was implemented to increase the model's accuracy. In addition to the technical aspects, an LED display board was integrated to provide real-time data to oncoming vehicles at road curves. This display showcases information such as vehicle types, passage time, and the number of vehicles within the curvy road. The proposed Multi-Scale Detector model, a customized version of the You Only Look Once (YOLO) architecture, achieved a mean average precision (mAP) of 0.954 at a confidence interval of 0.5. This overall research provides an effective and accurate solution for optimizing vehicle tracking, enhancing transportation safety, and enabling a variety of practical applications in traffic management and monitoring. © 2024 IEEE.

关键词： DeepSORT Euclidean Distance and Vehicle Intelligence Multi-scale Detector Object Detection Speed Tacking

来源：评论

学校读者我要写书评

暂无评论

Effective Pest Detection by YOLOv5 with BottleneckCSP 2

Effective Pest Detection by YOLOv5 with BottleneckCSP

引用

2nd International Conference on Power, Electrical, Electronics and Industrial Applications, PEEIACON 2024

作者： Tani, Ishrat Zahan Jannat, Tasmia Rajshahi University of Engineering & Technology Dept. of Computer Science & Engineering Rajshahi Bangladesh

ISBN: (纸本)9798331517984

Pest damage poses a significant risk to crop quality and productivity, underscoring the necessity for accurate and efficient detection methods for farmers. Artificial intelligence (AI) facilitates targeted pest management with reduced reliance on broad-spectrum insecticides and enables early detection of insect infestations. Object detection, a computer vision approach, is revolutionizing pest identification in agriculture by locating pests in digital crop photographs. However, the detection of tiny objects within these datasets presents challenges due to sparse spatial information. In response to this challenge, we have replaced the C3 block of YOLOv5 with the BottleneckCSP module, which is more effective for feature mapping of tiny objects. Our enhanced model demonstrates superior performance on the IP102 dataset, which is characterized by the presence of tiny objects. Comparative analysis reveals that our BottleneckCSP-integrated model significantly improves feature mapping, resulting in a notable increase in mean average precision (mAP). Specifically, our modified model achieves an mAP of 63.9%, outperforming many state-of-the-art methods on this dataset. This advancement in the YOLOv5 architecture sets a new benchmark for AI-powered precision agriculture, providing farmers with a highly efficient tool for crop protection and pest management. The significant performance gains highlight the potential of our approach to transform current agricultural practices, paving the way for more sustainable and effective pest control strategies. © 2024 IEEE.

关键词： Damage detection

来源：评论

学校读者我要写书评

暂无评论

MeQANet: Deep Learning-based Framework for Classifying Medical Query Answering in Bengali

MeQANet: Deep Learning-based Framework for Classifying Medic...

引用

2024 IEEE International Conference on Computing, Applications and Systems, COMPAS 2024

作者： Tasnia, Fairooz Hoque, Mohammed Moshiul Chittagong University of Engineering and Technology Dept. of Computer Science and Engineering Chattagram Bangladesh

ISBN: (纸本)9798331529765

The automatic classification of medical QA pairs is crucial for improving healthcare efficiency, allocating specialized care, optimizing resource distribution, and facilitating robust data analysis. Bengali, the 7th most widely spoken language globally, has a growing volume of health-related queries. However, unlike English, there is no benchmark large-scale medical QA dataset, pre-trained language model, or human baseline score for medical QA in Bengali. This paper introduces a deep learning-based framework called MeQANet that leverages CNN and BiLSTM techniques for classifying medical QA texts in Bengali. It also presents a novel medical QA corpus in Bengali (BnMedQA), consisting of approximately 4500 medical QA pairs across eight domains: Orthopedic, ENT, Gastroenterology and Hepatology, Gynecology, Pediatrics, General Medicine, Dermatology, and Cardiology. Various techniques (including LR, SVM, RF, MNB, AdaBoost (AB), CNN, BiL-STM, and CNN+BiLSTM) are employed to perform the classification tasks. Experimental results indicate that the CNN+BiLSTM approach achieved the highest F1-score of 0.85 among all employed techniques. © 2024 IEEE.

关键词： Query languages

来源：评论

学校读者我要写书评

暂无评论

Deep Feature Fusion Based Brain Tumor Classification From Imbalance MRI Dataset Using Weighted Loss Functions 2

Deep Feature Fusion Based Brain Tumor Classification From Im...

引用

2nd International Conference on Power, Electrical, Electronics and Industrial Applications, PEEIACON 2024

作者： Tasnim, Taieba Parvin, Farjana Rajshahi University of Engineering & Technology Dept. of Computer Science & Engineering Rajshahi Bangladesh

ISBN: (纸本)9798331517984

Deep learning using MRI images to classify brain tumours presents a significant challenge due to imbalanced training datasets, where one tumor type may have far more samples than others. The overrepresented class may perform worse in classifiers as a result of this imbalance. In order to tackle this issue, the paper examines how feature learning in CNNs is affected by weighted loss functions to manage data imbalance. The article propose deep feature fusion, where particular CNN model features that were trained using three distinct loss functions are fused and classified using softmax classifier. This approach lead to notable improvements in brain tumor classification compared to a CNN trained with standard cross-entropy loss. Notably, the proposed methods substantially reduce classification errors between uneven class samples. The proposed model achieved an impressive 94.5% classification accuracy, accompanied by an average Fl-score of 94.6%, average precision of 94.6 %, and average recall of 95%. These findings underscore the effectiveness of our methodology in improving brain tumor classification accuracy, offering a promising avenue for more reliable diagnosis and treatment planning in clinical settings, © 2024 IEEE.

关键词： Deep learning

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：