检索结果-内蒙古大学图书馆

Advancements and challenges of deep learning architectures for aerial image analysis: A systematic review

Intelligent systems with Applications 2025年 27卷

作者： Hashibul Ahsan shoaib Hadiur Rahman Nabil Md Anisur Rahman Md Mohsin Kabir M.F. Mridha Jungpil shin Information Technology St. Francis College Brooklyn NY 11201 United States Department of Computer Science and Engineering American International University-Bangladesh Dhaka 1229 Bangladesh School of Engineering and Technology Western Illinois University Illinois 61455 United States Division of Computer Science and Software Engineering Mälardalens University 722 20 Västerås Sweden School of Computer Science and Engineering The University of Aizu Aizu-wakamatsu 965-8580 Japan

The rapid advancement of deep learning (DL) technologies has significantly transformed the domain of aerial image analysis. This systematic review explores the forefront of deep learning architectures specifically designed for the processing and analysis of aerial imagery. It offers a comprehensive examination of updated models, such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Generative Adversarial Networks (GANs), and Transformers, highlighting their unique contributions and comparative effectiveness in aerial image analysis. This review critically compares these architectures through an extensive literature survey, focusing on their impact on enhancing accuracy, computational efficiency, and overall performance in critical aerial imaging tasks, such as classification, object detection, and semantic segmentation. Additionally, it sheds light on the innovative architectural improvements that have been crucial in overcoming traditional challenges associated with aerial image processing, such as handling high-resolution data, managing diverse and changing landscapes, and ensuring real-time analysis capabilities. by synthesizing current findings and identifying prevailing trends, this review not only charts the progress in the field but also outlines future research directions, emphasizing the need for more adaptable, robust, and efficient deep-learning solutions to meet the growing demands of aerial image analysis.

关键词： Deep learning Aerial image Object detection Image segmentation Image classification computer vision

来源：评论

学校读者我要写书评

暂无评论

A generative framework for detection and classification of plant leaf disease using diffusion network

引用

Applied soft Computing 2025年 177卷

作者： Das, Aryan Mahto, Rajul Wang, Wentao Jamali, Ali singh, Pravendra Roy, swalpa Kumar School of Computer Science Engineering Vellore Institute of Technology Bhopal Madhya Pradesh 466114 India Department of Computer Science University of Alabama at Birmingham 10th Ave S 35294 AL United States Department of Geography Simon Fraser University 8888 University Dr Burnaby BCV5A 1S6 Canada Department of Computer Science and Engineering Indian Institute of Technology Roorkee Uttarakhand 247667 India Department of Computer Science and Engineering Alipurduar Government Engineering and Management College West Bengal 736206 India

Agriculture is a crucial driver of global economic growth, significantly contributing to employment, Gross Domestic Product (GDP), and the provision of essential food and raw materials. However, crop diseases, which often go unnoticed and untreated, cause substantial yearly reductions in farm productivity. While effective on small-scale farms, traditional manual methods for disease detection are impractical for larger operations due to their inefficiency, subjectivity, and vulnerability to human error. In contrast, deep learning technologies offer a promising solution by automating the accurate recognition and classification of crop diseases, thus enhancing efficiency and accuracy without requiring highly specialized expertise. This work presents a novel approach by introducing a diffusion model for classifying plant leaf diseases. We propose a novel diffusion-based deep learning technique (LeafDisDiff), which employs a U-Net-style architecture with position embeddings, group normalization, attention blocks, and integrated residual blocks, centring around a denoising approach for each time step. The proposed LeafDisDiff algorithm has shown superior classification accuracy across three crop disease data benchmarks: Plant village, bangladeshi Crop Disease, and Apple Disease. For example, LeafDisDiff outperforms several deep learning classifiers such as viT, Cross viT, vGG-19, CNN, ResNet-50, and EfficientNet-b3 and improves classification accuracy by approximately 9.2, 6.2, 5.7, 5.5, 2.5, and 0.7 percentage points, respectively, achieving an average accuracy of 99.49% over the Plant village dataset. This work is motivated by recent advancements in diffusion probabilistic algorithms for generative image modelling and demonstrates the potential of these techniques in agricultural applications. Moreover, the implementation of this model can facilitate early and accurate detection of plant diseases, enabling prompt interventions like targeted pesticide application and sel

关键词： Crops

来源：评论

学校读者我要写书评

暂无评论

Refining adversarial attacks on machine-learning phishing webpage detectors through functionality-preserving HTML manipulations

引用

AIP Conference Proceedings 2024年第1期3193卷

作者： R. sonia s. Anila s. A. Karthik s. Meenakshi Eric Howard Kartikeya Parmar J. K. Periasamy v. vijayan Department of Computer Applications B S Abdur Rahman Crescent Institute of Science and Technology Vandalur Chennai Tamil Nadu 600048 India Department of Computer Science and Engineering Adi Shankara Institute of Engineering and Technology Ernakulam Kerala 683574 India Department of Information Science and Engineering BMS Institute of Technology and Management Bengaluru Karnataka 560064 India Department of S&H (General Engineering) R.M.K. Engineering College Kavarapettai Tamil Nadu 601 206 India Department of Physics and Astronomy Macquarie University Sydney Australia Department of Mechanical Engineering Nowgong Engineering College Madhya Pradesh 471201 India Department of Computer Science and Engineering Sri Sai Ram Engineering College Chennai Tamil Nadu 600044 India Department of Mechanical Engineering K.Ramakrishnan College of Technology Trichy Tamilnadu India

Research has demonstrated that Machine-Learning Phishing Webpage Detectors, which is vulnerable to malicious changes of the input webpage’s HTML code. Unfortunately, the newly suggested assaults have not been very effective so far because they target specific HTML components and don’t optimise the employment of the manipulations that have been accepted. To get over these limitations, we created a novel set of granular manipulations that can change input phishing webpage HTML code while keeping its aesthetic appeal and malevolent intent (alterations are rendered-preserving and functional by design). After that, we use a query-efficient black-box optimization technique to choose which alterations to apply in order to evade the target detector. In contrast to the weaker attacks created in earlier work, our results demonstrate that our attacks may completely demolish the performance of state-of-the-art ML-PWD with only 30 questions. This allows for a considerably more equitable evaluation of ML-robustness. PWD’s

关键词：

来源：评论

学校读者我要写书评

暂无评论

Predicting Dislocation Density in Martensite Ab-Initio

SSRN

引用

ssRN 2022年

作者： Andric, Predrag Echeverri Restrepo, sebastián Maresca, Francesco SKF B.V. Meidoornkade 14 Houten3992 AE Netherlands Department of Physics King’s College London Strand LondonWC2R 2LS United Kingdom Faculty of Science and Engineering University of Groningen Groningen9747AG Netherlands

The high dislocation density (ρd) in Fe-C and/or Fe-N martensite is known to be one of the main causes of its overall yield strength. Experiments indicate that the dislocation density and, therefore, the yield strength of martensite in low and medium C steels, follows a well defined behavior with respect to the background concentration of C (co). In the present work, we introduce a model based on the finding that the dislocation density after γ → α' is sufficiently high to accommodate/trap most of the available interstitials (C or N) within Cottrell atmospheres. We show that the number of interstitials trapped in Cottrell atmospheres scales with the interaction energy between the stress field of the dislocations and the C/N misfit tensor. The new model, which contains no free fitting parameters, shows that the dislocation density scales as ρd ∼ c01/4, leading to an extremely good match with experiments across a wide range of C and N concentrations. Therefore, this model sheds light on the mechanistic origin of dislocation strengthening in Fe-C and Fe-N martensite. © 2022, The Authors. All rights reserved.

关键词： Martensite

来源：评论

学校读者我要写书评

暂无评论

The Emerging Role of the Knowledge Driven Applications of Wireless Networks for Next Generation Online stream Processing

The Emerging Role of the Knowledge Driven Applications of Wi...

引用

International Conference on system Modeling & Advancement in Research Trends (sMART)

作者： vijaya Gunturu P. Rajani Kumari s. M. Chithra bhargabjyoti saikia Rajesh singh Devesh Pratap singh Department of ECE SR University Warangal Telangana India D.V.R. & Dr. H.S. MIC College of Technology Department of Mathematics R. M. K. College of Engineering and Technology Department of ECE DUIET Dibrugarh University Assam Uttaranchal Institute of Technology Uttaranchal University Department of Computer Science & Engineering Graphic Era Deemed to be University Dehradun Uttarakhand India

ISBN: (纸本)9781665487351

The present article discusses the use of stream processing to gather data from large-scale WIFI networks. Along with the foundational techniques for deliberate sampling, data collecting, likewise network monitoring in wireless networks, we also examine how understanding extraction may be viewed as an ML problem for applications for large-scale data streaming. We highlight the major This article discusses advancements in large data stream processing methods. We also look more closely at the database collection, edge detection, and methods for machine learning that may be used in the context of WIFI analytics. We discuss challenges, academic research, and the results of wireless network monitoring and stream analysis. Further research is anticipated into other dataflow improvements, such as pattern recognition and optimization algorithms.

关键词： Knowledge engineering Machine learning algorithms Wireless networks Image edge detection Machine learning Market research Wireless fidelity

来源：评论

学校读者我要写书评

暂无评论

Malware Detection Using Machine Learning Models 23rd

Malware Detection Using Machine Learning Models

引用

23rd International Conference on Network-based Information systems, Nbis 2020

作者： Carvalho, Glaucio H. s. Woungang, Isaac Anpalagan, Alagan Traore, Issa barolli, Leonard Sheridan College 1430 Trafalgar Rd OakvilleONL6H 2L1 Canada Department of Computer Science Ryerson University 350 Victoria Street TorontoONM5B 2K3 Canada Department of Electrical and Computer Engineering Ryerson University TorontoONM5B 2K3 Canada Department of Electrical and Computer Engineering University of Victoria P.O. Box 1700 STN CSC VictoriaBCV8W 2Y2 Canada 3-30-1 Wajiro-higashi Higashi-kuFukuoka811-0295 Japan

ISBN: (纸本)9783030578107

The criminal mind of adversaries is the most dangerous threat to enterprises and organizations worldwide. The creation and dissemination of malware across the Internet have given attackers of all levels an edge when it comes to compromise modern computer systems. Among the most promising tools to defeat nefarious initiatives, machine learning algorithms, and more specifically classifiers, arise as key techniques to detect malware infection across computer systems and consequently immunize the system at early stages of the propagation. In this paper, we present a performance comparison between neural networks and logistic regression for the problem of malware detection. Results show that neural networks and logistic regression are able to detect a malware presence considering the memory usage and the CPU usage of healthy and infected computers with an accuracy of more than 90%. © 2021, springer Nature switzerland AG.

关键词： Malware

来源：评论

学校读者我要写书评

暂无评论

Efficient Method for Personality Prediction using Hybrid Method of Convolutional Neural Network and LsTM

Efficient Method for Personality Prediction using Hybrid Met...

引用

Electronics and sustainable Communication systems (ICEsC), 2020 International Conference on

作者： N sujatha s Pramod shruti bhatla T Thulasimani Ravi Kant Amit Chauhan Department of Mathematics B. M. S. College of Engineering Bengaluru Karnataka India Department of Mathematics Nitte Meenakshi Institute of Technology Yelahanka Bangalore Karnataka India Department of Computer Science and Engineering Graphic Era Hill University Dehradun India Department of Mathematics Bannari Amman Institute of Technology Sathyamangalam Erode Tamilnadu India CSE Department Shoolini University Solan Himachal Pradesh India Department of life Sciences CHRIST (Deemed to be University) Bengaluru Karnataka India

Users’ contributions and the emotions conveyed in status updates may prove invaluable to studies of human behavior and character. A number of other research have taken a similar approach, and the field itself is still growing. The goal of this proposed is to create a technique for deducing a user’s personality traits based on their social media profiles. Among the many customer services now available on sNss are media and recommendations of user involvement. The need to give internet users with more specialized and customized services that meet their specific requirements, which sometimes depend heavily on the users’ inner personalities, is significant. However, there hasn’t been much work done on the psychological analysis that’s needed to deduce the user’s inner nature from their outward activities. In this instance, LsTM-CNN was fed pre-processed and vectorized text documents. sNF is used for feature extraction. The proposed method employs CFs for the purpose of Feature selection. Finally, LsTM-CNN was used to train the model. While CNN is good at extracting features that are independent of time, LsTM is better at capturing long-term dependencies. combination of features for personality prediction, the LsTM-CNN model is superior to the individual models.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Heart Disease Prediction using Logistic Regression with PCA-MFsA Feature Extraction Technique and Multidimensional scaling (MDs) Pre-processing Approach 3rd

Heart Disease Prediction using Logistic Regression with PCA-...

引用

3rd International Conference on Advancements in smart Computing and Information security, AsCIs 2024

作者： Priscila, s. silvia sakthivanitha, M. sadhana, C. Lavanya, M. Poornima, v. Department of Computer Science Bharath Institute of Higher Education and Research Tamil Nadu Chennai India Department of Information Technology Vels Institute of Science Technology and Advanced Studies Tamil Nadu Chennai India Department of MCA Prathyusha Engineering College Tamil Nadu Chennai India Department of Computer Applications S.D.N.B Vaishnav College for Women Tamil Nadu Chennai India Department of Computer Science Vels Isntitute of Science Technology and Advanced Studies Tamil Nadu Chennai India

ISBN: (纸本)9783031862922

Heart disease prediction is critical because it has the potential to save lives, increase patient outcomes, and save the cost of healthcare. An important task in medical diagnostics is the forecasting of heart disease. In this article, we investigate the efficacy of integrating Multidimensional scaling (MDs) preprocessing with various feature extraction methods for precise LR (Logistic Regression) heart disease prediction. The feature extraction methods under investigation include Independent Component Analysis (ICA), Principal Component Analysis (PCA) and Mean Fisher score-based Algorithm (PCA-MFsA). PCA with MFsA emerges as the most effective method, consistently producing superior outcomes in the forms of accuracy rate, precision, and recall value compared to ICA and PCA. This underscores the significance of combining PCA with the specialized feature selection process offered by PCA-MFsA to enhance the LR model’s discriminatory power. Moreover, The incorporation of MDs preprocessing with PCA-MFsA results in a significant improvement in accuracy rate, precision value, and recall. From the result obtained PCA-MFsA produces accuracy of 91%, precision of 0.90 and recall of 0.86 respectively. © The Author(s), under exclusive license to springer Nature switzerland AG 2025.

关键词： Logistic regression

来源：评论

学校读者我要写书评

暂无评论

Artificial intelligence for abnormality detection in high volume neuroimaging: a systematic review and meta-analysis

arXiv

引用

arXiv 2024年

作者： Agarwal, siddharth Wood, David Grzeda, Mariusz suresh, Chandhini Din, Munaib Cole, James Modat, Marc booth, Thomas C. School of Biomedical Engineering & Imaging Sciences King's College London Rayne Institute 4th Floor Lambeth Wing LondonSE1 7EH United Kingdom Leicester Medical School University of Leicester LeicesterLE1 7RH United Kingdom Centre for Medical Image Computing Department of Computer Science University College London LondonWC1V 6LJ United Kingdom Department of Neuroradiology King's College Hospital NHS Foundation Trust Ruskin Wing LondonSE5 9RS United Kingdom

Purpose: Most studies evaluating artificial intelligence (AI) models that detect abnormalities in neuroimaging are either tested on unrepresentative patient cohorts or are insufficiently well-validated, leading to poor generalisability to real-world tasks. The aim was to determine the diagnostic test accuracy and summarise the evidence supporting the use of AI models performing first-line, high-volume neuroimaging tasks. Methods: Medline, Embase, Cochrane library and Web of science were searched until september 2021 for studies that temporally or externally validated AI capable of detecting abnormalities in first-line CT or MR neuroimaging. A bivariate random-effects model was used for meta-analysis where appropriate. PROsPERO: CRD42021269563. Results: Only 16 studies were eligible for inclusion. Included studies were not compromised by unrepresentative datasets or inadequate validation methodology. Direct comparison with radiologists was available in 4/16 studies. 15/16 had a high risk of bias. Meta-analysis was only suitable for intracranial haemorrhage detection in CT imaging (10/16 studies), where AI systems had a pooled sensitivity and specificity 0·90 (95% CI 0·85-0·94) and 0·90 (95% CI 0·83-0·95) respectively. Other AI studies using CT and MRI detected target conditions other than haemorrhage (2/16), or multiple target conditions (4/16). Only 3/16 studies implemented AI in clinical pathways, either for pre-read triage or as post-read discrepancy identifiers. Conclusion: The paucity of eligible studies reflects that most abnormality detection AI studies were not adequately validated in representative clinical cohorts. The few studies describing how abnormality detection AI could impact patients and clinicians did not explore the full ramifications of clinical implementation. © 2024, CC bY.

关键词： Neurosurgery

来源：评论

学校读者我要写书评

暂无评论

Deep feature fusion and optimized feature selection based ensemble classification of liver lesions

引用

The Journal of Photographic science 2023年第6期71卷

作者： A. Anisha G. Jiji T. Ajith bosco Raj a Department of Computer Science and Engineering St. Xavier’s Catholic College of Engineering Nagercoil Tamil Nadu India b Department of Electronics and Communication Engineering Lord Jegannath College of Engineering and Technology Nagercoil Tamil Nadu India c Department of Electronics and Communication Engineering PSN College of Engineering and Technology Tirunelveli Tamil Nadu India

AbsTRACTClassification of liver abnormalities is crucial for the early identification of liver cancer. In clinical settings, radiological professionals typically make diagnoses manually which is subjective, time-consuming and vulnerable to error. Therefore, there is still a demand for precise classification of liver diseases. We propose an Ensemble learning-based classification model to classify liver lesions on CT images. To excerpt all the essential facts from the image, deep feature fusion is incorporated by concatenating the features from pre-trained deep CNN models densenet201 and InceptionResnetv2. To minimize the feature space and boost classification accuracy, hybrid optimization methodologies, Genetic Algorithms and Ant Colony Optimization are applied. Finally, a Heterogeneous Ensemble classifier divides the retrieved features into four groups (liver abscess, liver cirrhosis, hepatocellular carcinoma, and metastasis). It is clearly seen and observed that 98.3% accuracy is contributed by ensemble classifier with the support of concatenated deep features and this classifier excels in all other ways and means.

关键词： Liver Lesion Classification Ensemble Learning Feature Fusion Ant Colony Optimization Genetic Algorithms 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：