检索结果-内蒙古大学图书馆

Enhancing face recognition performance: a comprehensive evaluation of deep learning models and a novel ensemble approach with hyperparameter tuning

引用

Soft Computing 2024年第20期28卷 12399-12424页

作者： Selvaganesan, Jana Sudharani, B. Shekhar, S. N. Chandra Vaishnavi, K. Priyadarsini, K. Raju, K. Srujan Rao, T. Srinivasa Department of Electronics and Communication Engineering Vel Tech Rangarajan Dr. Sagunthala R & ampD Institute of Science and Technology Avadi Tamil Nadu Chennai600062 India Guntur Dist. Andhra Pradesh Vadlamudi India Department of Electronics and Communication Engineering Sreenidhi Institute of Science and Technology Telangana Hyderabad501301 India Department of Computer Science and Engineering Sona College of Technology Tamil Nadu Salem636005 India Department of Data Science and Business Systems School of Computing Faculty of Engineering and Technology SRM Institute of Science and Technology Tamil Nadu Kattankulathur603203 India Department of Computer Science and Engineering CMR Technical Campus Medchal Telangana Hyderabad501401 India Department of Electronics and Communication Engineering Aditya Engineering College Surampalem533437 India

In response to growing security concerns and the increasing demand for face recognition (FR) technology in various sectors, this research explores the application of deep learning techniques, specifically pre-trained Convolutional Neural Network (CNN) models, in the field of FR. The study harnesses the power of five pre-trained CNN models—DenseNet201, ResNet152V2, MobileNetV2, SeResNeXt, and Xception—for robust feature extraction, followed by SoftMax classification. A novel weighted average ensemble model, meticulously optimized through a grid search technique, is introduced to augment feature extraction and classification efficacy. Emphasizing the significance of robust data pre-processing, encompassing resizing, data augmentation, splitting, and normalization, the research endeavors to fortify the reliability of FR systems. Methodologically, the study systematically investigates hyperparameters across deep learning models, fine-tuning network depth, learning rate, activation functions, and optimization methods. Comprehensive evaluations unfold across diverse datasets to discern the effectiveness of the proposed models. Key contributions of this work encompass the utilization of pre-trained CNN models for feature extraction, extensive evaluation across multiple datasets, the introduction of a weighted average ensemble model, emphasis on robust data pre-processing, systematic hyperparameter tuning, and the utilization of comprehensive evaluation metrics. The results, meticulously analyzed, unveil the superior performance of the proposed method, consistently outshining alternative models across pivotal metrics, including Recall, Precision, F1 Score, Matthews Correlation Coefficient (MCC), and Accuracy. Notably, the proposed method attains an exceptional accuracy of 99.48% on the labeled faces in the wild (LFW) dataset, surpassing erstwhile state-of-the-art benchmarks. This research represents a significant stride in FR technology, furnishing a dependable and accurate

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

MapsTP: HD Map Images Based Multimodal Trajectory Prediction for Automated Vehicles 26

MapsTP: HD Map Images Based Multimodal Trajectory Prediction...

引用

26th Irish Machine Vision and Image Processing Conference, IMVIP 2024

作者： Sharma, Sushil Das, Arindam Sistu, Ganesh Halton, Mark Eising, Ciarán Department of Electronic & Computer Engineering University of Limerick Ireland SFI CRT Foundations in Data Science University of Limerick Ireland

ISBN: (纸本)9781837242672

Predicting ego vehicle trajectories remains a critical challenge, especially in urban and dense areas due to the unpredictable behaviours of other vehicles and pedestrians. Multimodal trajectory prediction enhances decision-making by considering multiple possible future trajectories based on diverse sources of environmental data. In this approach, we leverage ResNet-50 to extract image features from high-definition map data and use IMU sensor data to calculate speed, acceleration, and yaw rate. A temporal probabilistic network is employed to compute potential trajectories, selecting the most accurate and highly probable trajectory paths. This method integrates HD map data to improve the robustness and reliability of trajectory predictions for autonomous vehicles. © This is an open access article published by the IET under the Creative Commons Attribution License (http://***/licenses/by/3.0/)

关键词： Autonomous Vehicles HDMap Images MultiModel Trajectory Prediction Probabilistic Network

来源：评论

学校读者我要写书评

暂无评论

XGCN:a library for large-scale graph neural network recommendations

引用

Frontiers of computer science 2024年第3期18卷 247-249页

作者： Xiran SONG Hong HUANG Jianxun LIAN Hai JIN National Engineering Research Center for Big Data Technology and System Services Computing Technology and System LabCluster and Grid Computing LabSchool of Computer Science and TechnologyHuazhong University of Science and TechnologyWuhan 430074China Microsoft Research Asia Beijing 100080China

1 Introduction Graph Neural Networks(GNNs)have gained widespread adoption in recommendation systems,and nowadays there is a pressing need to effectively manage large-scale graph data[1].When it comes to large graphs,GNNs may encounter the scalability issue stemming from their multi-layer messagepassing ***,scaling GNNs has emerged as a crucial research area in recent years,with numerous scaling strategies being proposed.

关键词： scaling gained operations

来源：评论

学校读者我要写书评

暂无评论

Extracting Valuable Insights from The Handwritten Feedback 7

Extracting Valuable Insights from The Handwritten Feedback

引用

7th IEEE Pune Section International Conference, PuneCon 2024

作者： Chawale, Yugandhar Patwardhan, Prasanna Dewani, Rahul More, Ankit Jadhav, Varsha Pandey, Mrityunjoy Sharma, Neha Vishwakarma Institute of Information Technology Department of Artificial Intelligence and Data Science Maharashtra Pune India Army Institute of Technology Department of Computer Engineering Maharashtra Pune India Society for Data Science Pune India

ISBN: (纸本)9798331527822

This paper describes the collection and analysis of feedback data which is crucial for evaluating the success of any event and identifying areas of improvement. However, when feedback forms are distributed in paper format, extracting the text responses into a machine-readable format for analysis can be a tedious and error-prone manual process. Optical Character Recognition (OCR) Technology offers an automated solution for digitizing text from scanned documents, but its accuracy can be impacted by various factors like image quality, font styles, and language complexities. This study investigates the use of Gemini 1.5 Pro, a multimodal AI, to extract text data from PDF scans of collected handwritten feedback forms. After evaluation of Gemini's performance across a range of PDF qualities and comparing its output to manually transcribed data, the results demonstrate Gemini's high accuracy rates, even on lower quality scans, enabling efficient extraction of feedback responses while minimizing manual effort. This solution offers a scalable and reliable approach for educational institutions, professional organizations and other bodies to digitize survey responses for deeper qualitative and quantitative computational techniques. © 2024 IEEE.

关键词： data accuracy

来源：评论

学校读者我要写书评

暂无评论

Enhancing Carbon Emission Tracking with Gemini: An AI Solution for Sustainable Development 1

Enhancing Carbon Emission Tracking with Gemini: An AI Soluti...

引用

1st International Conference on AIML-Applications for engineering and Technology, ICAET 2025

作者： Dash, Annada Mathur, Bhavini Patil, Piyusha B.Tech in Computer Engineering Honors in Data Science and ML Cummins College of Engineering for Women Pune India B.Tech in Computer Engineering Cummins College of Engineering for Women Pune India

ISBN: (纸本)9798350355611

This research explores the integration of advanced AI technologies, specifically Google's Gemini-1.5-Flash-001 model and Vertex AI, into the development of EcoTrack-a mobile application that helps users monitor and reduce their carbon emissions. At the core of this innovation is EcoBot, an AI-powered chatbot embedded within EcoTrack, designed to deliver personalized, actionable insights on sustainability. By leveraging Gemini's capabilities alongside real-time data through Vertex AI, EcoBot enhances user engagement by offering tailored advice on reducing their carbon footprint. The study illustrates how EcoBot processes various data points, including user input and environmental metrics, to generate precise recommendations for users looking to make more sustainable choices. The focus of this work is on the importance of scalable AI frameworks that can adapt to evolving sustainability goals while efficiently managing real-time data. Additionally, this paper addresses the ethical dimensions of using AI for environmental purposes, discussing its potential to drive meaningful action at both the individual and societal levels, contributing to the global effort to combat climate change. © 2025 IEEE.

关键词： Carbon sequestration

来源：评论

学校读者我要写书评

暂无评论

Parasitized Malarial Infected Cell Detection through Variance Scaling Kernel Initializer Conv2D 4

Parasitized Malarial Infected Cell Detection through Varianc...

引用

4th International Conference on Advances in Electrical, Computing, Communication and Sustainable Technologies, ICAECT 2024

作者： Devi, M. Shyamala Umanandhini, D. Joshi, Aparna Dontamsetti, Gowtham Shreyas, S. Vel Tech Rangarajan Dr. Sagunthala R and D Institute of Science and Technology Computer Science and Engineering Tamilnadu Chennai India Pimpri Chinchwad College of Engineering Computer Science and Engineering Pune India Vel Tech Rangarajan Dr. Sagunthala R and D Institute of Science and Technology Artificial Intelligence and Data Science Tamilnadu Chennai India

ISBN: (纸本)9798350343670

Medical pathologists physically examine Plasmodium falciparum parasites infected blood cells that have been infected through a magnifying glass as part of the traditional diagnostic process, which is time-consuming and liable to mistakes. CNN and advanced technology have recently been used to conduct research on optimizing this procedure to identify orientation and geometrical information from blood sample particle images. With this view, this research recommends Variance Scaling Kernel Initializer Conv2D (VSKIC) Sequential model for predicting the infected malaria cell. The Malaria Cell Images dataset from KAGGLE, which contains 27,558 visuals of plasmodium parasites has been used for execution in the suggested VSKIC model. For the purpose of predicting a malarial cell infection, images are preprocessed and fitted using the VSKIC model. Three convolutional layers merged with max pooling layer, and three dense layers that are weight initialized with variance scaling kernel initializer are used as preprocessing steps on the malaria cell images for the purpose of identifying the presence of affected malarial tissue. The picture files were separated into training and evaluation images, and the training Malaria Cell images were used to evaluate the effectiveness of the VSKIC model along with the standard conV2D paradigms. While comparing the suggested model with the available CNN methods, the execution observations made with Python on a V100 Nvidia GPU workstation using 64 and 250 training intervals demonstrates that the proposed VSKIC model predicts with the higher accuracy of 98.96% and Area under curve of 97.8%. © 2024 IEEE.

关键词： accuracy AUC ConV2D initializer kernel

来源：评论

学校读者我要写书评

暂无评论

Leveraging XAI and Breakthrough Machine Learning Techniques for Trigeminal Neuralgia Severity Classification

Leveraging XAI and Breakthrough Machine Learning Techniques ...

引用

2024 IEEE Region 10 Symposium, TENSYMP 2024

作者： Abhijna, M.B. Akul, P. Kodipalli, Ashwini Global Academy of Technology Dept. of Information Science and Engineering India Global Academy of Technology Dept. of Computer Science and Engineering India Global Academy of Technology Dept. of Artificial Intelligence and Data Science India

ISBN: (纸本)9798350364866

Trigeminal Neuralgia (TN) is a debilitating chronic pain disorder that significantly diminishes overall well-being, making diagnosis and therapy more challenging. The quick and precise categorization of TN severity is critical to therapy effectiveness. To assess the severity of TN, this study makes use of advanced machine learning techniques and explainable AI (XAI) approaches. We use a range of bagging and boosting strategies, including AdaBoost, Random Forest, Decision Tree, K-Nearest Neighbours (KNN), Gradient Boosting, and Logistic Regression. Random Forest was 95.2% accurate, whereas AdaBoost and Gradient Boosting were 92% and 97.2% in terms of their accuracy. The Decision Tree classifier obtained an accuracy of 84.62%, while K-Nearest Neighbours and Logistic Regression also achieved 92.31% accuracy. F1- score, precision, and recall metrics were used to assess performance on a dataset that included patient demographics and medical histories. XAI techniques, such as LIME and SHAP values, were used to improve the models' readability and transparency, which helped to progress the creation of cutting-edge diagnostic tools and increase clinician and researcher confidence. © 2024 IEEE.

关键词： Logistic regression

来源：评论

学校读者我要写书评

暂无评论

A Combination Model of Periodic and Non-periodic Events for Temporal Knowledge Graph Event Prediction

引用

Chinese Journal of Electronics 2025年第3期34卷 952-961页

作者： School of Computer Science and Information Security Guilin University of Electronic Technology Guilin 541004 China School of Data Science and Software Engineering Wuzhou University Wuzhou 543002 China

Temporal knowledge graph (TKG) reasoning aims to predict missing facts or future events at given timestamps and has attracted more and more attention in recent years. Existing TKG reasoning methods mainly focus on the interactions between entities and ignore the associations between events where the entities involve. In addition, the characteristics of different types of events have not been studied and exploited, which reduces the performance of event prediction. To address these problems, this paper proposes a combination model of periodic and non-periodic events (CM-PNP). Specifically, there are two basic components designed to process different types of events. The periodic component of CM-PNP learns the recurrent pattern of periodic events and encodes the temporal information in the manner of timespan to prevent the unseen timestamp issue. The non-periodic component of CM-PNP introduces extra information (e.g., entity attributes) to represent non-periodic events, and predicts this type of events according to the related historical events. A combination model of multiple sub-models that focus on encoding different parts of the event is used to improve the performance of single model. The periodic and non-periodic components are combined by a gate block. The experimental results on three real-world datasets demonstrate that CM-PNP outperforms the existing baselines.

关键词： Temporal knowledge graph Temporal knowledge graph reasoning Event prediction Embedding learning data mining

来源：评论

学校读者我要写书评

暂无评论

Privacy Enhanced Mobile User Authentication Method Using Motion Sensors

引用

computer Modeling in engineering & sciences 2024年第3期138卷 3013-3032页

作者： Chunlin Xiong Zhengqiu Weng Jia Liu Liang Gu Fayez Alqahtani Amr Gafar Pradip Kumar Sharma Shenzhen Institutes of Advanced Technology Chinese Academy of SciencesShenzhen518052China Sangfor Technologies Inc. Shenzhen518055China School of Data Science and Artificial Intelligence Wenzhou University of TechnologyWenzhou325035China College of Computer Science&Technology Zhejiang University of TechnologyHangzhou310023China Software Engineering Department College of Computer and Information SciencesKing Saud UniversityRiyadh12372Saudi Arabia Math&Computer Science Department Faculty of ScienceMenofia UniversityShebin El-KomEgypt Computing Science Department University of AberdeenAberdeenUK

With the development of hardware devices and the upgrading of smartphones,a large number of users save privacy-related information in mobile devices,mainly smartphones,which puts forward higher demands on the protection of mobile users’privacy *** present,mobile user authenticationmethods based on humancomputer interaction have been extensively studied due to their advantages of high precision and non-perception,but there are still shortcomings such as low data collection efficiency,untrustworthy participating nodes,and lack of *** this end,this paper proposes a privacy-enhanced mobile user authentication method with motion sensors,which mainly includes:(1)Construct a smart contract-based private chain and federated learning to improve the data collection efficiency of mobile user authentication,reduce the probability of the model being bypassed by attackers,and reduce the overhead of data centralized processing and the risk of privacy leakage;(2)Use certificateless encryption to realize the authentication of the device to ensure the credibility of the client nodes participating in the calculation;(3)Combine Variational Mode Decomposition(VMD)and Long Short-TermMemory(LSTM)to analyze and model the motion sensor data of mobile devices to improve the accuracy of model *** experimental results on the real environment dataset of 1513 people show that themethod proposed in this paper can effectively resist poisoning attacks while ensuring the accuracy and efficiency of mobile user authentication.

关键词： Mobile authentication blockchain federated learning smart contract certificateless encryption VMD LSTM

来源：评论

学校读者我要写书评

暂无评论

Detecting Driver Fatigue with Python and OpenCV 10

Detecting Driver Fatigue with Python and OpenCV

引用

10th International Conference on Communication and Signal Processing, ICCSP 2024

作者： Pandi, C. Anish, T.P. Selvanayaki, S. Jagadeesan, P. Vedaraj, M. Sanjay, R. Vel Tech Multi Tech Dr.Rangarajan Dr.Sakunthala Engineering College Computer Science and Engineering Avadi India R.M.K College of Engineering and Technology Computer Science and Engineering Tamil nadu India Saveetha Engineering College Artificial Intelligence and Data Science Tamil nadu Thandalam India R.M.D Engineering College Computer Science and Engineering Tamil nadu India

ISBN: (纸本)9798350353068

Utilizing eye aspect ratio analysis and facial landmarks, the sleep detection tool is intended to detect indicators of sleepiness or drowsiness in people. By leveraging the face recognition library, the program detects and extracts facial landmarks, with a particular focus on locating the positions of the eyes. The software analyzes the eye aspect ratio (EAR), which takes into account the relative distances between particular eye landmarks, to ascertain if an individual has fully or partially closed their *** video frame capture is performed from a chosen video source, typically a webcam, allowing real-time monitoring of the user. Each frame is processed to detect and analyze the eye regions, computing the EAR values for both the left and right eyes. If the EAR falls below a predefined threshold for a consecutive number of frames, it indicates that the person's eyes are closed for an extended duration, suggesting potential drowsiness or *** enhance user safety, an alarm sound is played when the program detects drowsiness, providing an immediate alert to the individual. The alarm sound serves as an auditory cue, helping the person regain alertness and avoid potential accidents or undesired consequences. Additionally, the program displays the computed EAR on the video feed, allowing for visual feedback and monitoring of the eye aspect ratio values for debugging purposes and fine-tuning the system's *** employing this sleep detection program, various applications can benefit from its capabilities. For instance, it can be utilized in the context of drowsy driving detection systems, alerting drivers when they exhibit signs of fatigue and helping prevent accidents due to impaired driving. Moreover, the program holds potential for use in other scenarios where monitoring individuals' alertness and detecting sleep-related patterns is essential, such as workplace safety or medical *** summary, the sleep detection program employs faci

关键词： Video signal processing

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：