检索结果-内蒙古大学图书馆

Sufficient learning: mining denser high-quality pixel-level labels for edge detection

Neural Computing and Applications 2025年第14期37卷 8245-8260页

作者： Chen, Xiaodiao Yang, Wenya Wu, Wen Tao, Xiuting Mao, Xiaoyang School of Computer Science and Technology Hangzhou Dianzi University Hangzhou310018 China School of Media and Design Hangzhou Dianzi University Hangzhou310018 China Department of Computer Science and Engineering University of Yamanashi Kofu400-8511 Japan

The success of deep neural networks can largely be attributed to large-scale datasets with accurate annotations. In many practical applications, labels are annotated by multiple annotators, resulting in ambiguous labels. To mitigate the impact of such labels, existing deep edge detectors tend only to exploit sparsely reliable samples with high-confident labels by simple voting mechanism. However, lacking sufficient pixel-level learnable samples would severely degrade detection performance. To combat this issue, this work designs a sufficient learning framework for edge detection (SLED) to sufficiently explore reliable samples for model learning, which consists of three stages: sample selection, label refinement, and retraining. Firstly, we propose a sample selection scheme founded on history prediction, which aims to identify more reliable samples and locate unreliable regions containing potentially reliable samples. Secondly, we design an efficient label refinement strategy based on a graph convolutional network (GCN) to propagate reliable label information to unreliable nodes. This strategy can generate "new training targets" with denser, reliable labels for all images. Finally, we retrain the network with these new targets, obtaining a more robust model. Experimental results on three widely used benchmarks show that the proposed detector outperforms all compared SOTA methods. Our code will be released at https://***/wenya1994/SLED. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Pixels

来源：评论

学校读者我要写书评

暂无评论

Deep Learning-based Hybrid Recommendation System for Safe Approach to Curated Cosmetics 3

Deep Learning-based Hybrid Recommendation System for Safe Ap...

引用

3rd International Conference on Intelligent Data Communication Technologies and Internet of Things, IDCIoT 2025

作者： Rohitha Chowdary, G. Tharun, J. Sivaiah, G. Raveena, S. Surendran, R. Department of Computer Science and Engineering Saveetha School of Engineering Saveetha Insititute of Medical and Technical Science Chennai India

ISBN: (数字)9798331527549

ISBN: (纸本)9798331527549

In the rapidly evolving cosmetics market, it is imperative to prioritize customer safety while offering personalized recommendations. This study introduces a hybrid recommendation system for cosmetics, using deep learning to combine the advantages of collaborative filtering with content-based filtering. The system assesses user preferences, skin types, and ingredient sensitivities, recommending solutions that are both effective and safe. The system proposes an integration of user-based and item-based collaborative filtering methodologies with deep neural networks to effectively elucidate intricate links among individuals, items, and their characteristics. This hybrid technique facilitates the system's ability to propose goods that provide both aesthetic appeal and skin safety, therefore fostering a healthier and more confident lifestyle. The system has an impressive accuracy of 92.5% in recommending suitable cosmetic products for individual clients. This enhanced accuracy reduces the likelihood of negative reactions and elevates customer satisfaction via personalized suggestions. © 2025 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Intelligent surveillance support system

引用

Discover Internet of Things 2023年第1期3卷 9页

作者： Saketh, Meduri Nandal, Neha Tanwar, Rohit Reddy, B. Pradyun Department of Computer Science Engineering Gokaraju Rangaraju Institute of Engineering and Technology Hyderabad India Geethanjali College of Engineering and Technology Hyderabad India School of Computer Science University of Petroleum and Energy Studies Dehradun India Department of Computer Science Engineering Sreenidhi Institute of Engineering and Technology Hyderabad India

The Intelligent Surveillance Support System(ISSS) is an innovative software solution that enables real-time monitoring and analysis of security footage to detect and identify potential threats. This system incorporates advanced features such as face recognition, alarm on theft detection, visitors in/out detection and motion detection, to provide a comprehensive and reliable security solution. The implementation of this software aims to improve the efficiency of surveillance systems, thereby enhancing the safety and security of public and private spaces. The focus of this study is on performing the aforementioned tasks in real time while utilizing enhanced algorithms from the OpenCV Library, such as LBPH and Haar Cascading, which enhance the use of machine perception and help us produce outcomes with an accuracy of about 95% after multiple runs. With the rapid advancements in technology and the increasing need for surveillance in today’s world, the Intelligent Surveillance Support System holds immense potential in the field of security and surveillance. © The Author(s) 2023.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

Intelligent Underwater Sound Surveillance for Intrusion Detection and Emergency Alerting 15

Intelligent Underwater Sound Surveillance for Intrusion Dete...

引用

15th International Conference on Computing Communication and Networking Technologies, ICCCNT 2024

作者： Mohite, Vaishnavi Shivaji Patra, Krishna Gayatri Sankar, Ch Hari Srinivas, M. Ullas, S. Amrita School of Engineering Department of Computer Science and Engineering Bengaluru India

ISBN: (纸本)9798350370249

With the demand of protective necessity of underwater environment and infrastructures, it is high time for the technological advancement towards intelligently monitoring and ensuring responsive security mechanism. A novel approach uses modern signal analysis and machine learning methods for the differentiation and subsequent analysis of underwater sounds based on the specific source, focusing on interfering objects like submarines, ships, or other noise caused by humans. The introduced system entails the use of the real-time monitoring system so as to allow constant assessment of the acoustic data collected beneath the water. In case of exposure to any abnormal sounds, an alarm system is activated meant to inform a monitoring and control centre. Its versatility for various underwater conditions is addressed by exercising of the system that considers the fluctuations of water temperature, pressure, and acoustic background. The state centralized monitoring of the system makes it easy for the system to respond to the potential threats, faster than using the graphical user interface. The following sections of this paper describe both the general system design and the architecture and feature extraction of the proposed machine learning algorithms, specific experimental test results and where appropriate demonstrations of the specific features and an assessment of the envisioned role of this system in underwater security and emergency alerting. © 2024 IEEE.

关键词： Emergency Alerting Machine Learning for Underwater Surveillance Underwater Acoustic Audio Processing

来源：评论

学校读者我要写书评

暂无评论

Inspires effective alternatives to backpropagation:predictive coding helps understand and build learning

引用

Neural Regeneration Research 2025年第11期20卷 3215-3216页

作者： Zhenghua Xu Miao Yu Yuhang Song School of Health Sciences and Biomedical Engineering Hebei University of TechnologyTianjinChina Department of Computer Science University of OxfordOxfordUK

Artificial neural networks are capable of machine learning by simulating the hiera rchical structure of the human *** enable learning by brain and machine,it is essential to accurately identify and correct the prediction errors,referred to as credit assignment(Lillicrap et al.,2020).It is critical to develop artificial intelligence by understanding how the brain deals with credit assignment in neuroscience.

关键词： assignment learning enable

来源：评论

学校读者我要写书评

暂无评论

Mutual Retinex: Combining Transformer and CNN for Image Enhancement

引用

IEEE Transactions on Emerging Topics in Computational Intelligence 2024年第3期8卷 2240-2252页

作者： Jiang, Kui Wang, Qiong An, Zhaoyi Wang, Zheng Zhang, Cong Lin, Chia-Wen Harbin Institute of Technology School of Computer Science and Technology Harbin150001 China Wuhan University National Engineering Research Center for Multimedia Software School of Computer Science Wuhan430072 China Wuhan Polytechnic University School of Mathematics and Computer Science Wuhan430024 China National Tsing Hua University Department of Electrical Engineering Institute of Communications Engineering Hsinchu30013 Taiwan

Images captured in low-light or underwater environments are often accompanied by significant degradation, which can negatively impact the quality and performance of downstream tasks. While convolutional neural networks (CNNs) and Transformer architectures have made significant progress in computer vision tasks, there are few efforts to harmonize them into a more concise framework for enhancing such images. To this end, this study proposes to aggregate the individual capability of self-attention (SA) and CNNs for accurate perturbation removal while preserving background contents. Based on this, we carry forward a Retinex-based framework, dubbed as Mutual Retinex, where a two-branch structure is designed to characterize the specific knowledge of reflectance and illumination components while removing the perturbation. To maximize its potential, Mutual Retinex is equipped with a new mutual learning mechanism, involving an elaborately designed mutual representation module (MRM). In MRM, the complementary information between reflectance and illumination components are encoded and used to refine each other. Through the complementary learning via the mutual representation, the enhanced results generated by our model exhibit superior color consistency and naturalness. Extensive experiments have shown the significant superiority of our mutual learning based method over thirteen competitors on the low-light task and ten methods on the underwater image enhancement task. In particular, our proposed Mutual Retinex respectively surpasses the state-of-the-art method MIRNet-v2 by 0.90 dB and 2.46 dB in PSNR on the LOL 1000 and FIVEK datasets, while with only 19.8% model parameters. © 2017 IEEE.

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

Pruning techniques for artificial intelligence networks: a deeper look at their engineering design and bias: the first review of its kind

引用

Multimedia Tools and Applications 2025年第11期84卷 9591-9665页

作者： Mohanty, Lopamudra Kumar, Ashish Mehta, Vivek Agarwal, Mohit Suri, Jasjit S. School of Computer Science Engineering and Technology Bennett University Gr. Noida India Department of Computer Science ABES Engineering College UP Ghaziabad India Stroke Diagnostic and Monitoring Division AtheroPoint™ RosevilleCA United States

Trained Artificial Intelligence (AI) models are challenging to install on edge devices as they are low in memory and computational power. Pruned AI (PAI) models are therefore needed with minimal degradation in performance. This is the first study of its kind where we have characterized three types of PAI paradigms, namely, weight pruning (WP), channel pruning (CP), and hybrid pruning (HP). Further, the study also analyses the risk of bias (RoB) in PAI models, and explainable PAI (XPAI) networks. PRISMA model was adapted for the selection of the best 145 studies. The review had two hypotheses: (i) PAI had bias and (ii) the order of performance for the three PAI is: HP > CP > WP in terms of performance. Thirty-two AI attributes using each of two experts were categorized into four clusters, namely, architecture, optimization, dataset design, and scientific validation. The RoB was analyzed in these PAI studies using AP(ai)Bias 2.0 (AtheroPoint™, CA, USA), which was based on a scoring-and-ranking strategy. This was then compared against the Butterfly model, composed of radial and regional biases, and analyzed using a Venn diagram in three independent bins based on low-moderate and moderate-high cut-offs. Recommendations were laid out for low-bias studies. PAI studies that lack adequate AI attributes were more inclined to the RoB. It has been observed that 19.64%, 21.42%, and 58.92% studies were in low-bias, moderate-bias, and high-bias bins, respectively. Pruning made inference faster and reduced sparsity. We have observed low participation of the PAI studies in the XPAI framework. The study presented the challenges and five-step recommendations for mitigating bias in PAI models. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Risk assessment

来源：评论

学校读者我要写书评

暂无评论

A Real Time Prediction Model for Round Winners of Counter Strike:Global Offensive 3

A Real Time Prediction Model for Round Winners of Counter St...

引用

3rd International Conference on Sentiment Analysis and Deep Learning, ICSADL 2024

作者： Paul, Rosebell Sajeer, Rihan Vaishnav, M.V. Abraham, Savio Vinu Sahajan, Shreya SCMS School of Engineering and Technology Department of Computer Science and Engineering India

ISBN: (纸本)9798350396157

This research study aims to create a full time winner predictor program for Counter-Strike Global Offensive tournaments that can predict the round winner of the game being spectated. The algorithm creates accurate predictions of the round winner. The data required for the prediction has to be extracted from the game servers. The data is collected from the game servers using an API. The prediction can be viewed live by the user with the help of a GUI. The project can provide CS:GO enthusiasts and bettors with real-time insights and predictions for round winners in competitive matches. System aims to enhance the user's understanding of match dynamics and support informed decision making. © 2024 IEEE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

A novel and efficient statistical and soft-computing intelligence integrated feature selection technique for human chronic diseases prediction

引用

Multimedia Tools and Applications 2025年 1-44页

作者： Yadav, Amit Khanna, Munish Anand, Darpan Department of Computer and Engineering Sir Padampat Singhania University Udaipur Bhatewar India School of Computer Science and Engineering Galgotias University Yamuna Expressway Greater Noida203201 India

Due to the exponential increase in data volume, the widespread use of intelligent information systems has created significant obstacles and issues. High dimensionality and the existence of noisy and extraneous data are a few of the difficulties. These difficulties incur high computing costs and have a considerable effect on the accuracy and efficiency of machine learning (ML) methods. A key idea used to increase classification accuracy and lower computational costs is feature selection (FS). Finding the ideal collection of features that can accurately determine class labels by removing unnecessary data is the fundamental goal of FS. However, finding an effective FS strategy is a difficult task that has given rise to a number of algorithms built using biological systems based soft computing approaches. In order to solve the difficulties faced during the FS process;this work provides a novel hybrid optimization approach that combines statistical and soft-computing intelligence. On the first dataset of diabetes disease, the suggested approach was initially tested. The approach was later tested on the Wisconsin Diagnostic Breast Cancer (WDBC) dataset after yielding encouraging results on diabetes dataset. While finding the solution, typically, data cleaning happens at the pre-processing stage. Later on, in a series of trials, different FS methods were used separately and in hybridized fashion, such as fine-tuned statistical methods like lasso (L1 regularization) and chi-square, as well as binary Harmony search algorithm (HSA) which is based on soft computing algorithmic approach. The most efficient strategy was chosen based on the performance metric data. These FS methods pick informative features, which are then used as input for a variety of traditional ML classifiers. The chosen technique is shown along with the determined influential features and associated metric values. The success of the classifiers is then evaluated using performance metrics like accuracy, preci

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

Deep Fake Analyser: A Review Based on Detecting the Deepfakes 1

引用

3rd International Conference on Artificial-Business Analytics, Quantum and Machine Learning: Trends, Perspectives, and Prospects, Com-IT-Con 2023

作者： Srivastava, Santosh Kumar Srivastava, Durgesh Kantha, Praveen Mahto, Manoj Kumar Kumar, Sheo Singh, Hare Ram Department of Computer Science and Engineering School of Computing Science and Engineering and Technology Galgotias University Greater Noida India Department of Computer Science and Engineering Chitkara University Institute of Engineering and Technology Chitkara University Punjab Rajpura India Department of Computer Science and Engineering School of Engineering and Technology Chitkara University Baddi Himachal Pradesh Baddi India Department of Computer Science and Engineering Vignan Institute of Technology and Science Hyderabad India Department of Computer Science and Engineering CMR Engineering College Hyderabad India Department of Computer Science and Engineering GNIOT Greater Noida India

ISBN: (数字)9789819725083

ISBN: (纸本)9789819725076

With the speedy growth in the technology and automation sectors, different techniques have been developed which can easily manipulate multimedia content such as videos and images with the ultimate level of realism. It becomes very hard for a person to find out which of the media content is real and which is fake. With the help of deep learning and GANs (Generative Adversarial Networks), deep fakes can be produced. A Deep fake is a fake video or image where a person in an existing video or image is replaced by someone else by using an algorithm. If we look at the positives, we can see how deep fake technology can be used in advertisements, filmmaking, and video games. But on the contrary, it also results in the spread of fake news and the generation of pornographic content. This seriously impacts society and presents enormous threats. This research paper aims to present a paper based on a deep fake analyzer that will be able to identify and recognize fake media with the help of deep neural networks. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Generative adversarial networks

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：