检索结果-内蒙古大学图书馆

Thoracic-net: Explainable artificial intelligence (XAI) based few shots learning feature fusion technique for multi-classifying thoracic diseases using medical imaging

引用

Multimedia Tools and Applications 2025年第9期84卷 5397-5433页

作者： Bhosale, Yogesh H. Patnaik, K. Sridhar Zanwar, S.R. Singh, Saket Kr. Singh, Vandana Shinde, U.B. Department of Computer Science & Engineering Birla Institute of Technology Ranchi Mesra India India India India

Background & Need: The early detection of thoracic diseases and COVID-19 (coronavirus disease) significantly limits propagation and increases therapeutic outcomes. This article focuses on swiftly distinguishing COVID-19 patients with 10 chronic thoracic illnesses from healthy examples. The death rates of COVID-19-confirmed patients are rising due to chronic thoracic illnesses. Method: To identify thoracic illnesses (Consolidation, Tuberculosis, Edema, Fibrosis, Hernia, Mass, Nodule, Plural-thickening, Pneumonia, Healthy) from X-ray images with COVID-19, we provide an ensemble-feature-fusion (FFT) deep learning (DL) model. 14,400 chest X-ray images (CXRI) of COVID-19 and other thoracic illnesses were obtained from five public sources and applied UNet-based data augmentation. High-quality images were intended to be provided under the CXR standard. To provide model parameters and feature extractors, four deep convolutional neural networks (CNNs) with a proprietary CapsNet as the backbone were employed. To generate the ensemble-fusion classifiers, we suggested five additional USweA (Unified Stacking weighted Averaging)-based comparative ensemble models as an alternative to depending solely on the findings of the single base model. Additionally, USweA enhanced the models' performance and reduced the base error-rate. USweA models were knowledgeable of the principles of multiple DL evaluations on distinct labels. Results: The results demonstrated that the feature-fusion strategy performed better than the standalone DL models in terms of overall classification effectiveness. According to study results, Thoracic-Net significantly improves COVID-19 context recognition for thoracic infections. It achieves superior results to existing CNNs, with a 99.75% accuracy, 97.89% precision, 98.69% recall, 98.27% F1-score, shallow 28 CXR zero-one loss, 99.27% ROC-AUC-score, 1.45% error rate, 0.9838 MCC, (0.98001, 0.99076) 95% CI, and 5.708 s to test individual CXR. This suggested USweA m

关键词： COVID-19

来源：评论

学校读者我要写书评

暂无评论

Phishing Webpage Detection: Unveiling the Threat Landscape and Investigating Detection Techniques

引用

IEEE Communications Surveys and Tutorials 2025年第2期27卷 974-1007页

作者： Kulkarni, Aditya Balachandran, Vivek Das, Tamal Indian Institute of Technology Dharwad Department of Computer Science and Engineering Dharwad580011 India Singapore Institute of Technology InfoComm Technology 10 Dover Dr Singapore

In the realm of cybersecurity, phishing stands as a prevalent cyber attack, where attackers employ various tactics to deceive users into gathering their sensitive information, potentially leading to identity theft or financial gain. Researchers have been actively working on advancing phishing webpage detection approaches to detect new phishing URLs, bolstering user protection. Nonetheless, the ever-evolving strategies employed by attackers, aimed at circumventing existing detection approaches and tools, present an ongoing challenge to the research community. This survey presents a systematic categorization of diverse phishing webpage detection approaches, encompassing URL-based, webpage content-based, third-party-based and visual techniques. Through a comprehensive review of these approaches and an in-depth analysis of existing literature, our study underscores current research gaps in phishing webpage detection. Furthermore, we suggest potential solutions to address some of these gaps, contributing valuable insights to the ongoing efforts to combat phishing attacks. © 1998-2012 IEEE.

关键词： Phishing

来源：评论

学校读者我要写书评

暂无评论

Bidirectional Legendre memory unit: bidirectional memory for person authentication combining voice and online signature

引用

Neural Computing and Applications 2025年第3期37卷 1541-1563页

作者： Kumar, Rohitesh Ghosh, Rajib Department of Computer Science and Engineering National Institute of Technology Patna Ashok Rajpath Patna800005 India

With the increasing popularity of smart portable electronic gadgets, voice-based online person verification systems have become prevalent. However, these systems are susceptible to attacks where illegitimate individuals exploit the recorded voices of legitimate users, leading to false confirmations—spoofing attacks. To overcome this limitation, this article presents an innovative solution by combining speech and online handwritten signatures to mitigate the risks associated with spoofing attacks in voice-based authentication systems because a person has to be present in front of the system to produce an online handwritten signature. To accomplish this objective, this work proposes a novel bidirectional Legendre memory unit (BLMU), a type of recurrent neural network (RNN), for person authentication (verification) and recognition. The Legendre memory unit (LMU) is an innovative memory cell for RNNs that efficiently retains temporal/non-temporal sequential information over a long period with minimal resources. It achieves information orthogonalization by solving coupled ordinary differential equations (ODEs) and leveraging Legendre polynomials, ensuring effective data representation. The proposed framework for person authentication and recognition comprises seven convolution layers, four BLMU layers, two dense layers, and one output layer. The performance of the proposed BLMU-based deep learning framework has been evaluated on a self-generated/private dataset of combined feature matrix of voice signals and online handwritten signatures in the Devanagari script. To assess performance, experiments have also been conducted using various RNN architectures, such as LSTM, BLSTM, and ordinary differential equation recurrent neural network (ODE-RNN), to have a performance comparison with the proposed BLMU-based deep learning (DL) framework. The results demonstrate the superiority of the proposed BLMU-based DL framework in enhancing the accuracy of person verification systems,

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

Automated breast nuclei feature extraction for segmentation in histopathology images using Deep-CNN-based gaussian mixture model and color optimization technique

引用

Multimedia Tools and Applications 2025年 1-27页

作者： Murmu, Anita Kumar, Piyush Computer Science and Engineering National Institute of Technology Patna Ashok Rajpath Bihar Patna800005 India

Hematoxylin and eosin (H&E) staining are the key sources for identifying breast cancer patterns with different colors and shapes of nuclei cells for segmenting histopathology nucleus images. In nucleus cells, the color categorization of hematoxylin is stained blue, whereas cytoplasmic cells of eosin are stained pink. This color dissimilarity in the nucleus cells indicates different structures of the cellular and cancer-specific cells, which need to detect on time is a challenge for accurate diagnosis. To overcome this issue, a Deep Convolution Neural Network (Deep-CNN) based Unet framework with Transfer Learning (TL) is proposed for segmenting histopathology breast cancer images. Specifically, the Deep-CNN framework has been used to extract features automatically from input images, and select the features by using the TL technique for minimizing the target tasks to train it faster. Furthermore, a deep convolutional-based Gaussian Mixture Model (GMM) is used to normalize the shapes and merge the nucleus cell pixels for color optimization. The proposed method has been tested on Kaggle’s 2018 Data science Blow histopathology breast image dataset. The experimental results of the proposed scheme have been evaluated by performance evaluation matrices, namely accuracy, Jaccard coefficient, Dice Similarity Coefficient (DSC), Intersection over Union (IoU), and F1-score. The proposed approach outperforms compared with different state-of-the-art models on the same dataset. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Verification of news and certificate, and plagiarism detector

引用

Multimedia Tools and Applications 2025年 1-28页

作者： Jain, Pushpak Anand, Dewansh Rajni Behal, Vansh Kapoor, Vipul Sharma, Neha Jindal, Neeru Department of Electronics and Communication Engineering Thapar Institute of Engineering and Technology Punjab Patiala India Department of Computer Science and Engineering Chitkara University Institute of Engineering and Technology Chitkara University Punjab Patiala India

Fake news, Fake certification, and Plagiarism are the most common issues arising these days. During this COVID-19 situation, there are a lot of rumors and fake news spreading and some of us are using fake certification to avoid vaccination and plagiarism became the most common topic during online studies. It is found that several apps and websites are present for the detection of false news and plagiarism. However, there is no single website for the detection of fake certificates. Moreover, nothing is made that consists of all these features in one platform. So, this proposed work develops a website for detecting fake news, Fake certificate, and plagiarism detector. The proposed method uses Natural Language Processing techniques for detecting NPC (news, plagiarism, and certificate). The proposed work uses image processing with some Python libraries like OpenCV in all the detections and that makes it quite interesting. For detecting NPC (news, plagiarism, and certificate) the algorithm is built using a k-mean model, logistic regression, and neural network. The proposed work uses the datasets that are trained using the count vectorizer method for the detection and to get its accuracy which will be tested using machine learning algorithms. The proposed method has shown an accuracy rate of 94.28% on Logistics Regression in comparison to existing methods. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Fake news Machine learning Natural language processing Plagiarism Social media platform

来源：评论

学校读者我要写书评

暂无评论

A moment-based pooling approach in convolutional neural networks for breast cancer histopathology image classification

引用

Neural Computing and Applications 2025年第2期37卷 1127-1156页

作者： Kumar, Arvind Singh, Chandan Sachan, Manoj Kumar Department of Computer Science and Engineering Sant Longowal Institute of Engineering and Technology Longowal Punjab Sangrur148106 India Department of Computer Science Punjabi University Punjab Patiala147002 India

Automated analysis of breast cancer (BC) histopathology images is a challenging task due to the high resolution, multiple magnifications, color variations, the presence of image artifacts, and morphological variability in the tissue images, often resulting in misdiagnosis. We propose four CNN models using a novel Zernike moments (ZMs)-based pooling approach, which replaces the existing global average pooling for the effective classification, referred to as CZP-CNN, MZP-CNN, SPMZP-CNN, and SPCZP-CNN. We utilize the ZMs-based pooling at three intermediate and last convolution layers. Of the four models, the two models, SPMZP-CNN and SPCZP-CNN, employ spatial pyramidal structures of feature maps to represent global and local characteristics of the feature maps. They effectively integrate the ZMs of various orders derived from the pyramidal representation of a feature map, representing the global and local aspects of shape and fine morphological details. The ability of ZMs is fully utilized to capture shape and morphological information available in the feature map using only a few low-order moments, which adds to the robustness of the proposed approach. Detailed experimental results on the publicly available BreaKHis dataset with four magnification levels and on the high-resolution BACH dataset demonstrate the proposed models’ effectiveness, robustness, and generalization ability. Among these proposed models, the complex ZMs-based SPCZP-CNN model achieves an impressive mean classification accuracy of 94.16% on the BreaKHis dataset and overall best binary and multiclass classification accuracy of 96.25% and 93.75% on the BACH dataset. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

BGRU-MTRA: bilinear GRU networks with multi-path temporal residual attention for suspicious activity recognition

引用

Neural Computing and Applications 2025年第1期37卷 185-212页

作者： Pandey, Ajeet Kumar, Piyush Computer Science and Engineering National Institute of Technology Patna Ashok Rajpath Bihar Patna800005 India

Suspicious activity recognition (SAR) is an active research field in computer vision and image processing due to the rapid demand for intelligent video surveillance systems. However, current automated systems focus too much on the temporal dynamics of events in videos and ignore the importance of spatial dynamics. The existing methods for SAR are complex and demand substantial resources. Hence, this paper proposes a novel trade-off architecture for SAR integrating the strengths of the multi-path temporal residual attention (MTRA) module with the bilinear-gated recurrent unit (BGRU) module. The MTRA module combines spatial feature extraction (SFE) with temporal residual attention network (TRAN) blocks for improving SAR by enhancing resilience to variations in object sizes, viewpoints, and motion patterns. It selects relevant action features, addresses vanishing gradient issues, and reduces spatial dimension. The BGRU module preserves spatial features, improving the model’s ability to recognize fine-grained features and complex temporal patterns for effectively recognizing suspicious activities. The BGRU-MTRA system yields recognition accuracy of 93.28%, 98.97%, 99.86%, and 99.66%, and 48.42% on the benchmark hybrid crime action (HCA), UT interaction (UTI), CAVIER, hockey fight (HF), and the most challenging UCF-crime datasets, respectively. Through comprehensive experiments, it is clear that the proposed method outperforms state-of-the-art methods over benchmark datasets with reduced parameters. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Spatio-temporal data

来源：评论

学校读者我要写书评

暂无评论

Robust Detection of Malicious Encrypted Traffic via Contrastive Learning

引用

IEEE Transactions on Information Forensics and Security 2025年 20卷 4228-4242页

作者： Shen, Meng Wu, Jinhe Ye, Ke Xu, Ke Xiong, Gang Zhu, Liehuang Beijing Institute of Technology School of Cyberspace Science and Technology Beijing100081 China Beijing Institute of Technology School of Computer Science Beijing100081 China Tsinghua University Department of Computer Science and Technology Beijing100084 China Institute of Information Engineering Chinese Academy of Sciences Beijing100190 China

Traffic encryption is widely used to protect communication privacy but is increasingly exploited by attackers to conceal malicious activities. Existing malicious encrypted traffic detection methods rely on large amounts of labeled samples for training, limiting their ability to quickly respond to new attacks. These methods also are vulnerable to traffic obfuscation strategies, such as injecting dummy packets. In this paper, we propose SmartDetector, a robust malicious encrypted traffic detection method via contrastive learning. We first propose a novel traffic representation named Semantic Attribute Matrix (SAM), which can effectively distinguish between malicious and benign traffic. We also design a data augmentation method to generate diverse traffic samples, which makes the detection model more robust against different traffic obfuscation strategies. We propose a malicious encrypted traffic classifier that first pre-trains a model via contrastive learning to learn deep representations from unlabeled data, then fine-tunes the model with a supervised classifier to achieve accurate detection even with only a few labeled samples. We conduct extensive experiments with five public datasets to evaluate the performance of SmartDetector. The results demonstrate that it outperforms the state-of-the-art (SOTA) methods in three typical scenarios. Specifically, in the evasion attack detection scenario, SmartDetector achieves an F1 score and AUC above 93%, with average improvements of 19.84% and 18.17% over the SOTA method, respectively. © 2005-2012 IEEE.

关键词： Data obfuscation

来源：评论

学校读者我要写书评

暂无评论

Multi-View Test-Time Adaptation for Semantic Segmentation in Clinical Cataract Surgery

引用

IEEE Transactions on Medical Imaging 2025年第5期44卷 2307-2318页

作者： Li, Heng Ou, Mingyang Li, Haojin Qiu, Zhongxi Niu, Ke Fu, Huazhu Liu, Jiang Southern University of Science and Technology Research Institute of Trustworthy Autonomous Systems Shenzhen518055 China Southern University of Science and Technology Department of Computer Science and Engineering Shenzhen518055 China Beijing Information Science and Technology University Computer School Beijing100192 China Institute of High Performance Computing Agency for Science Technology and Research Fusionopolis Way 138632 Singapore Southern University of Science and Technology Research Institute of Trustworthy Autonomous Systems Department of Computer Science and Engineering Shenzhen518055 China

Cataract surgery, a widely performed operation worldwide, is incorporating semantic segmentation to advance computer-assisted intervention. However, the tissue appearance and illumination in cataract surgery often differ among clinical centers, intensifying the issue of domain shifts. While domain adaptation offers remedies to the shifts, the necessity for data centralization raises additional privacy concerns. To overcome these challenges, we propose a Multi-view Test-time Adaptation algorithm (MUTA) to segment cataract surgical scenes, which leverages multi-view learning to enhance model training within the source domain and model adaptation within the target domain. In the training phase, the segmentation model is equipped with multi-view decoders to boost its robustness against variations in cataract surgery. During the inference phase, test-time adaptation is implemented using multi-view knowledge distillation, enabling model updates in clinics without data centralization or privacy concerns. We conducted experiments in a simulated cross-center scenario using several cataract surgery datasets to evaluate the effectiveness of MUTA. Through comparisons and investigations, we have validated that MUTA effectively learns a robust source model and adapts the model to target data during the practical inference phase. © 1982-2012 IEEE.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Adaptive Attention Mechanisms Based Multicriteria Recommender Systems for Improved User Preference Discovery

引用

SN computer science 2025年第4期6卷 1-16页

作者： Rajput, Ishwari Singh Tewari, Anand Shanker Tiwari, Arvind Kumar Department of Computer Science & Engineering National Institute of Technology Patna Bihar Patna India Department of Computer Science & Engineering Kamla Nehru Institute of Technology Sultanpur (UP) India

Personalized recommender systems are becoming more popular to reduce the issue of information overload. It is also observed that the recommendations provided by multi-criteria recommender system (MCRS) are more accurate and precise in comparison to conventional single-criteria recommender systems. MCRS incorporates customers’ preferences based on multiple criteria while recommending any item. Recent research has shown that MCRS can efficiently use implicit and explicit input across various dimensions to make exact recommendations across multiple domains. Due to use of multiple criteria, existing MCRS methods often struggle with effectively aggregating and interpreting these criteria, making it hard to accurately determine user preferences. In this article, we introduce a novel adaptive attention mechanism that dynamically assigns weights to each criterion based on user preferences. Each criterion is initially weighted based on its importance to a user. An attention mechanism is used to understand the exact user’s preference for each criterion. Later, the nearest-neighbor method finds users with similar interest to predict unknown ratings which is further used for Top-N recommendations. The performance of the proposed approach (AM-MCRS) is validated on Yahoo! movies multi-criteria dataset having 1716 unique users and 965 different movies and a comparative analysis is performed with existing single and multi-criteria recommender systems. The results clearly show that the AM-MCRS methodology outperform baseline methods, achieving the highest precision (0.9412), F-score (0.9343), accuracy (0.929), and coverage (0.9407), while also observing a significant reduction in MAE (0.7623) and RMSE (0.9145). These findings highlight the capability of AM-MCRS to improve recommendation accuracy and dependability across various domains. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2025.

关键词： Attention mechanism Deep learning Dynamic user preference Multi-criteria recommender system Nearest-neighbor

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：