检索结果-内蒙古大学图书馆

Perception-Aware Attack Against Music Copyright Detection: Impacts and Defenses

IEEE Transactions on Dependable and Secure Computing 2025年第3期22卷 2959-2977页

作者： Duan, Rui Qu, Zhe Zhao, Shangqing Ding, Leah Liu, Yao Lu, Zhuo University of Missouri-Kansas City School of Science and Engineering Kansas CityMO64110 United States Central South University School of Computer Science and Engineering Changsha410017 China University of Oklahoma School of Computer Science TulsaOK74135 United States American University Department of Computer Science WashingtonDC20016 United States University of South Florida Department of Computer Science and Engineering TampaFL33620 United States University of South Florida Department of Electrical Engineering TampaFL33620 United States

Recently, adversarial machine learning attacks have posed serious security threats against practical audio signal classification systems, including speech recognition, speaker recognition, music copyright detection. Most existing studies have mainly fo cused on ensuring the effectiveness of attacking an audio signal classifier via creating a noise-like perturbation on the original signal, which remains a gap in preserving the human perception of adversarial audios. This paper presents a novel perspective to cre ate adversarial audios by integrating the human perception model into the attack formulation to generate well-perceived adversarial *** from conventional approaches which primarily focused on using Lp norm to preserve the audio quality, we adopt a human study to understand how human participants react to different types of music perturbations, build a Siamese Neural Network (SNN) based model to characterize the human percep tion. The new findings of the human perception study guide us to formulate a new computationally efficient, multiple-feature-based perception-aware (CEMF-PA)attack, which manipulates different audio signal features to find an optimal perturbed music signal against music copyright detection. This novel attack vector opens a new door to generating highly effective, well-perceived adversarial audio signals via manipulating the auditory features. Experimental results show that the proposed attack is effective against YouTube’s copyright detection. Finally, we propose the defense strategy design to make the copyright detection more robust to adversarial music signals generated by the CEMF-PA attack. © 2004-2012 IEEE.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

Explainable cost-sensitive deep neural networks for brain tumor detection from brain MRI images considering data imbalance

引用

Multimedia Tools and Applications 2025年 1-28页

作者： Shawon, Md Tanvir Rouf Shibli, G. M. Shahariar Ahmed, Farzad Joy, Sajib Kumar Saha Department of Computer Science and Engineering Ahsanullah University of Science and Technology Dhaka Bangladesh

This paper presents a research study on the use of Convolutional Neural Network (CNN), ResNet50, InceptionV3, EfficientNetB0 and NASNetMobile models to efficiently detect brain tumors in order to reduce the time required for manual review of the report and create an automated system for classifying brain tumors. An automated pipeline is proposed, which encompasses five models: CNN, ResNet50, InceptionV3, EfficientNetB0 and NASNetMobile. The performance of the proposed architecture is evaluated on a balanced dataset and found to yield an accuracy of 99.33% for fine-tuned InceptionV3 model. Furthermore, Explainable AI approaches are incorporated to visualize the model’s latent behavior in order to understand its black box behavior. To further optimize the training process, a cost-sensitive neural network approach has been proposed in order to work with imbalanced datasets which has achieved almost 4% more accuracy than the conventional models used in our experiments. The cost-sensitive InceptionV3 (CS-InceptionV3) and CNN (CS-CNN) show a promising accuracy of 92.31% and a recall value of 1.00 respectively on an imbalanced dataset. The proposed models have shown great potential in improving tumor detection accuracy and must be further developed for application in practical solutions. We have provided the datasets and made our implementations publicly available at -https://***/shahariar-shibli/Explainable-Cost-Sensitive-DeepNeural-Networks-for-Brain-Tumor-Detection-from-Brain-MRI-Images. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Adaptive pelican optimization with optimized mask RCNN for automatic lung cancer detection

引用

Multimedia Tools and Applications 2025年 1-21页

作者： Sudha, R. Maheswari, K.M. Uma Department of Computer Science and Engineering SRM Institute of Science and Technology Chengalpattu India Department of Computing Technologies SRM Institute of Science and Technology Chengalpattu India

Lung cancer is a prevalent and deadly disease worldwide, necessitating accurate and timely detection methods for effective treatment. Deep learning-based approaches have emerged as promising solutions for automated medical image analysis. This study proposes an enhanced Mask R-CNN framework tailored specifically for the automatic detection and severity analysis of lung cancer from CT images. The proposed approach consists of three stages namely, pre-processing, lung nodule detection, and segmentation using enhanced Mask R-CNN and severity analysis. Our framework employs a deep convolutional neural network architecture trained on a comprehensive dataset of annotated lung images. By incorporating a region-based convolutional neural network (R-CNN) with a mask prediction branch, our model accurately localizes lung tumors while providing precise pixel-level segmentation masks. To enhance the performance of mask RCNN, the parameter present in the classifier is optimally selected using the adaptive pelican optimization (APO) algorithm. The proposed framework detects lung tumors and provides a comprehensive severity analysis, enabling clinicians to assess cancer stage and progression accurately. Evaluation on a benchmark dataset demonstrates superior detection accuracy and robustness compared to existing methods. Our enhanced Mask R-CNN approach shows promise as a valuable tool for early diagnosis and severity assessment of lung cancer, potentially improving patient outcomes and healthcare efficiency. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Lung cancer

来源：评论

学校读者我要写书评

暂无评论

ROUGE-SciQFS: A ROUGE-based Method to Automatically Create Datasets for Scientific Query-Focused Summarization 31

ROUGE-SciQFS: A ROUGE-based Method to Automatically Create D...

引用

31st International Conference on Computational Linguistics, COLING 2025

作者： Ramirez-Orta, Juan Milios, Evangelos Maguitman, Ana Soto, Axel J. Faculty of Computer Science Dalhousie University Canada Institute for Computer Science and Engineering UNS - CONICET Department of Computer Science and Engineering Universidad Nacional del Sur Argentina

ISBN: (纸本)9798891761964

So far, the task of Scientific Query-Focused Summarization (Sci-QFS) has lagged in development when compared to other areas of Scientific Natural Language Processing because of the lack of data. In this work, we propose a methodology to take advantage of existing collections of academic papers to obtain large-scale datasets for this task automatically. After applying it to the papers from our reading group, we introduce a novel dataset for Sci-QFS composed of 8,695 examples, each one with a query, the sentences of the full text from a paper and the relevance labels for each. After testing several classical and state-of-the-art embedding models on this data, we found that the task of Sci-QFS is far from being solved, although it is relatively straightforward for humans. Surprisingly, we found that classical methods outperformed modern pre-trained Deep Language Models (sometimes by a large margin), showing the need for large datasets to better fine-tune the latter. We share our experiments, data and models at https://***/jarobyte91/rouge_sciqfs. © 2025 Association for Computational Linguistics.

关键词： Structured Query Language

来源：评论

学校读者我要写书评

暂无评论

uFOIL: An Unsupervised Fusion of Image Processing and Language Understanding

引用

IEEE Access 2025年 13卷 31683-31700页

作者： Rahman, Md. Abdur Hasan, Md. Tanzimul Farooq Howlader, Umar Kader, Md. Abdul Islam, Md. Motaharul Hung Pham, Phuoc Mehedi Hassan, Mohammad United International University Department of Computer Science and Engineering Dhaka1212 Bangladesh Providence College Department of Mathematics and Computer Science ProvidenceRI02918 United States King Saud University College of Computer and Information Sciences Department of Information Systems Riyadh11543 Saudi Arabia

In academic institutions, processing and evaluating documents such as exam scripts remains a labor-intensive process susceptible to human error. Traditional digitization systems face significant challenges in handling the complexities of mixed handwritten and printed text and varying document structures. These challenges are exacerbated by the absence of annotated datasets due to privacy concerns, particularly in contexts involving sensitive exam evaluations. To address these issues, this study introduces uFOIL, an unsupervised ensemble-based framework that integrates advanced image and language processing techniques to automate the extraction and validation of key information. The framework employs a majority voting mechanism that combines four state-of-the-art optical character recognition systems. Furthermore, a transformer architecture is incorporated to enhance contextual understanding and the structured formatting of extracted text that follows a post-processing confidence scoring mechanism. The proposed framework achieves high performance, with accuracies of 95.77% and 96.48% for student names and IDs, respectively;and 95.07% for total mark validation based on a dataset of exam script samples. Additionally, experiments on the benchmark ICDAR2013 dataset suggest the framework’s strong applicability achieving precision, recall, and F1-scores of 95.89%, 97.86%, and 96.87%, respectively. © 2025 The Authors.

关键词： Optical character recognition

来源：评论

学校读者我要写书评

暂无评论

Optimised brain tumour detection with the Pteropus Unicinctus optimisation reliant deep model

引用

International Journal of Intelligent Information and Database Systems 2025年第2期17卷 186-216页

作者： Chhabra, Sumit Bansal, Khushboo Desh Bhagat University Punjab Mandi Gobindgarh India Department of Computer Science Khalsa College for Women Amritsar India Department of Computer Science and Engineering Desh Bhagat University Punjab Mandi Gobindgarh India

This research Pteropus Unicinctus optimisation (PUO) is proposed which is mathematically derived from the echolocation character of Pteropusand and the eye power searching ability of Unicinctus. The integration of convolutional neural networks (CNNs) and long short-term memory (LSTM) in the PUO-reliant deep model presents an optimal solution to enhance the precision and dependability of brain tumour detection. This strategy leverages the unique capabilities of both architectures and fine-tunes their performance through hyperparameter optimisation, which leads to better diagnostic results and improved quality of patient care in brain tumour detection. The proposed approach attained the accuracy of 99.51%, 99.58%, and 99.50%, and considering sensitivity 99.48%, 99.60%, 99.37% attained, and the model attained 99.48%, 99.58%, and 99.26% of specificity for BraTS 2019 dataset, SimBRATS 2019 dataset, and brain tumour dataset respectively. Copyright © 2025 Inderscience Enterprises Ltd.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

On Neutrosophy logic for Multiple Correlation Analysis in Fuzzy Uncertain Database

引用

Neutrosophic Sets and Systems 2025年 76卷 1-17页

作者： Wazir, Samar Beg, M.M. Sufyan Ahmad, Tanvir Sohail, Shahab Saquib Department of Computer Engineering Jamia Millia Islamia New Delhi India Department of Computer Engineering Aligarh Muslim University Aligarh India School of Computing Science and Engineering VIT Bhopal University MP Sehore466114 India

The primary goal of this paper is to introduce a novel method for mining frequent and interesting items by incorporating correlation analysis between two items in an uncertain transactional database using the OWA operator. The work is further expanded by proposing two additional methods for mining frequent and interesting items, utilizing fuzzy means and the OWA operator along with multiple correlation analysis for more than two items in uncertain transactional databases. The effectiveness of the proposed methods is evaluated by running the algorithms on both standard and example datasets, with results compared to the traditional probabilistic approach for identifying frequent and interesting items through multiple correlation. While multiple correlation analysis highlights the relationships of interestingness and uninterestingness between items, the OWA operator enhances the results when combined with fuzzy means and probabilistic methods. Additionally, the paper suggests a future research direction using Neutrosophy logic, which is anticipated to open new avenues for further exploration in this field. © 2025, University of New Mexico. All rights reserved.

关键词： Database systems

来源：评论

学校读者我要写书评

暂无评论

Monitoring the crowd emotion using valence and arousal of crowd based on prominent features of crowd

引用

Signal, Image and Video Processing 2025年第7期19卷 1-12页

作者： Manojkumar, K. Suji Helen, L. Department of Computer Science and Engineering Sathyabama Institute of Science and Technology Chennai India

Analysing the crowd behaviour is a potential research component for ensuring societal safety by consistent video surveillance and management through officials. Public safety management is an inevitable process in developing and developed countries and possess a strong relationship with individual over a crowd behaviour. However, the challenges are doubled when the applications attempt to monitor or predict the crowd behaviours. The standard emotional cues utilized by the conventional machine learning algorithms highly rely on facial features, motion and gestures of the participating members in a crowd. The article concentrates on a novel approach to determine the crowd emotions through provocations captured due to emotional valence. The different features are combined in a fuzzy reasoning model to emphasize on the facial recognition of every individual in a crowd, along with other primary features of harmful substances or objects for measuring the public safety metrics. The common parameters considered from the crowd are magnitude, variance, confusion index, density of the crowd, along with provocations. Provocations of the crowd are measured based on the enthalpy metric and magnitude of the crowd. Density of the crowd, motion, trajectory and the confusion index are processed as the inputs of the reinforcement learning algorithm. The outcomes of the model described the probability of provocations and valence derived from the fuzzy reasoning of the crowd motion, emotions, and thus highlights the scenes with potentially harmful situations. The entire model was constructed using a DarkNet model in a Convolutional Neural Network with YOLOv8 and the simulation results have showcased that the proposed model outperforms the conventional approaches of crowd emotion sensing. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Detection of Quality Deep Fake Images and Videos Using Customised Convolutional Neural Networks 1

Detection of Quality Deep Fake Images and Videos Using Custo...

引用

1st International Conference on Advances in computer science, Electrical, Electronics, and Communication Technologies, CE2CT 2025

作者： Mahar, Alap Verma, Pushpneel Singh, Ajit Bhagwant University Department of Computer Science and Engineering Rajasthan Ajmer305004 India VMSB University Department of Computer Science and Engineering India

ISBN: (纸本)9798331518578

Due to advancements in artificial intelligence there are numerous deep fake images and video collections are available on the Internet and social media. The primary aim of the study is to analyse the deep fake images and videos since the quality is constantly improving a novel method was developed for accurate detection of quality deep fakes. The suggested approach uses a customised convolutional neural network (CNN) technique that uses facial landmark detection to extract structured data from photos and video frames before feeding it into the methods of CNN. The customised CNN model is augment-based CNN for generation of fake images and fake data. Involves about 260 films from the data set in which 202 images are made up while others were real images. About 300 videos are used in which 250 videos are fake and 50 were real the proposed model achieved the accuracy of 95.58% and 0.97 AUC score that outperforms the existing models like MLP-CNN and CNN. Additionally, the method succeeds with greater accuracy then the conventional models like DST -Net, VGG 16 Efficient Net. This research study's primary goal is to create a new CNN learning method for identifying high-quality deep fake photos and videos. © 2025 IEEE.

关键词： Video analysis

来源：评论

学校读者我要写书评

暂无评论

Disaster Management Systems: Utilizing YOLOv9 for Precise Monitoring of River Flood Flow Levels Using Video Surveillance

引用

SN computer science 2025年第3期6卷 1-14页

作者： Shankar, G. Geetha, M. Kalaiselvi Ezhumalai, P. Department of Computer Science and Engineering FEAT Annamalai University Chidambaram 608002 India Department of Computer Science and Engineering R.M.D. Engineering College Chennai 601206 India

Flood disasters pose significant threats to human lives and infrastructure, necessitating advanced methods for the timely and accurate monitoring of water levels in rivers. This study introduces an innovative approach utilizing the recent YOLOv9 (You Only Look Once, version 9) deep learning model to detect and monitor river water flow levels effectively. By leveraging YOLOv9’s robust object detection capabilities, the proposed system achieves precise segmentation and real-time analysis of river water regions. The method not only identifies river water areas with high accuracy but also quantifies flow levels, providing critical data for flood prediction and management. Comprehensive testing was conducted using diverse datasets representing various river conditions and flow scenarios. The results demonstrate that YOLOv9 significantly outperforms traditional methods in terms of detection speed and accuracy, making it a valuable tool for enhancing flood disaster response and management strategies. This work contributes to the development of more resilient and proactive flood disaster management systems through the integration of state-of-the-art machine learning techniques, offering significant improvements in predictive capabilities and response effectiveness. The proposed system was trained and evaluated using real-time river flood datasets, achieving a notable accuracy rate of 98%, a mean average precision (mAP) of 97.5%, precision of 98%, and recall of 98% in predicting flood level characteristics. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2025.

关键词： Deep learning Flood disaster management Machine learning Object detection River water flow detection Water level segmentation YOLOv9

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：