检索结果-内蒙古大学图书馆

CLIMS++: Cross Language Image Matching with Automatic Context Discovery for Weakly Supervised Semantic Segmentation

International Journal of computer Vision 2025年 1-20页

作者： Xie, Jinheng Deng, Songhe Hou, Xianxu Luo, Zhaochuan Shen, Linlin Huang, Yawen Zheng, Yefeng Shou, Mike Zheng Computer Vision Institute School of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China Department of Computer Science Wenzhou-Kean University Wenzhou325060 China Guangdong Provincial Key Laboratory of Intelligent Information Processing Shenzhen University Shenzhen China Show Lab National University of Singapore Singapore639798 Singapore School of AI and Advanced Computing Xi’an Jiaotong-Liverpool University Suzhou215400 China Jarvis Research Center Tencent YouTu Lab Shenzhen518703 China

While promising results have been achieved in weakly-supervised semantic segmentation (WSSS), limited supervision from image-level tags inevitably induces discriminative reliance and spurious relations between target classes and background regions. Thus, Class Activation Map (CAM) usually tends to activate discriminative object regions and falsely includes lots of class-related backgrounds. Without pixel-level supervisions, it could be very difficult to enlarge the foreground activation and suppress those false activation of background regions. In this paper, we propose a novel framework of Cross Language Image Matching with Automatic Context Discovery (CLIMS++), based on the recently introduced Contrastive Language-Image Pre-training (CLIP) model, for WSSS. The core idea of our framework is to introduce natural language supervision to activate more complete object regions and suppress class-related background regions in CAM. In particular, we design object, background region, and text label matching losses to guide the model to excite more reasonable object regions of each category. In addition, we propose to automatically find spurious relations between foreground categories and backgrounds, through which a background suppression loss is designed to suppress the activation of class-related backgrounds. The above designs enable the proposed CLIMS++ to generate a more complete and compact activation map for the target objects. Extensive experiments on PASCAL VOC 2012 and MS COCO 2014 datasets show that our CLIMS++ significantly outperforms the previous state-of-the-art methods. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

An Empirical Comparison of Machine and Deep Learning Algorithms for Predicting Maternal Health Risk 5th

An Empirical Comparison of Machine and Deep Learning Algorit...

引用

5th EAI International Conference on Cognitive Computing and Cyber Physical Systems, IC4S 2024

作者： N, Srinivasa Gupta Valarmathi, B. Kavitha, A. Gopini, Sai Bhuvan Seela, Hari Kumar Department of Manufacturing Engineering School of Mechanical Engineering Vellore Institute of Technology Tamil Nadu Vellore India Department of Software and Systems Engineering School of Computer Science Engineering and Information Systems Vellore Institute of Technology Tamil Nadu Vellore India Department of Mathematics School of Advanced of Sciences Vellore Institute of Technology Tamil Nadu Vellore India

ISBN: (纸本)9783031770777

Maternal health is among the greatest challenges in the world, especially in rural areas as there lack medical practitioners, they do not have easily accessible publics clinics and transport is difficult. Therefore, high rates of maternal as well as infant morbidity and mortalities are recorded. This research utilizes Artificial Intelligence (AI) with machine learning algorithms to forecast and address maternal health hazards right at their onset stage. The current research utilizes the concept of AI along with many Machine Learning (ML) methods like the Ensemble Learning Model (ELM), Random Forest (RF), K-Nearest Neighbour (KNN), Decision-Tree (DT), XG-Boost (XGB), Cat Boost (CB), and Gradient Boosting (GB), along with Synthetic Minority Over-sampling Technique (SMOTE) algorithm used for dealing with the problem class imbalance within the data set. SMOTE algorithm is utilized for the dataset balancing process. The handling system involves refining data preprocessing with the help of feature engineering and robust data cleaning which makes sure that anomalies do not erode the reliability of the predictive model. The existing methods [1] used RF (90%), DT (87%), XGB (85%), CB (86%), and GB (81%) algorithms and were compared with the accuracies of the proposed models like Logistic Regression (LR), Ensemble Learning Bagging (ELB), Ensemble Learning Stacking (ELS), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU). The existing methods used only imbalance dataset. The accuracies of the proposed models with using SMOTE algorithm (balanced dataset) are LR (61.33%), KNN (81%), ELB (92.33%), ELS (90.66%) CNN (40.67%), RNN (59.67%), LSTM (54%), GRU (56%) respectively. Among these methods, ELB achieved 92.33% of accuracy with using SMOTE algorithm using imbalanced dataset. Whereas the accuracies of the proposed models without using SMOTE algorithm (imbalanced dataset) are LR (66.09%), KNN (68.47%)

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Oral Cancer Classification Using a Hybrid Attention-Aided Deep Learning Model 5th

Oral Cancer Classification Using a Hybrid Attention-Aided D...

引用

5th International Conference on Medical Imaging and computer Aided Diagnosis, MICAD 2024

作者： Dutta, Srinjan Roy, Jibesh Mandal, Diptarka Kaplun, Dmitrii Shpakovskaya, Irina Sarkar, Ram Department of Computer Science and Engineering Jalpaiguri Government Engineering College Jalpaiguri India School of Computer Engineering and Mathematical Sciences Defence Institute of Advanced Technology Pune India Artificial Intelligence Research Institute China University of Mining and Technology Xuzhou China Department of Automation and Control Process Saint Petersburg Electrotechnical University "LETI" Saint Petersburg Russia Department of Computer Science and Engineering Jadavpur University West Bengal Kolkata India I.P. Pavlov Saint Petersburg First State Medical University St. Petersburg Russia

ISBN: (纸本)9789819638628

Oral cancer remains a critical global health challenge, characterized by high morbidity and mortality due to late-stage diagnosis. This paper addresses the need for improved diagnostic accuracy by introducing a novel hybrid-attention aided deep learning model for oral cancer classification. Our approach integrates the EfficientNetB0 model for feature extraction with a customized hybrid attention module, combining self, spatial, and soft attention mechanisms to emphasize crucial features. The model has been trained and evaluated on both microscopic and macroscopic oral cancer datasets, demonstrating its generalizability. Our proposed model achieves an accuracy of 96.97% on a macroscopic dataset and 92% on a microscopic dataset, with respective F1-scores of 0.97 and 0.89, outperforming baseline models. This hybrid attention-aided deep learning model shows significant potential in improving early detection and treatment planning of oral cancer. Source code and additional results of this work can be found at the GitHub repository. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Self-Learning Based Dependable Offloading Optimization in Semi-Trusted Vehicular Edge Computing and Networks

引用

IEEE Transactions on Vehicular Technology 2025年第5期74卷 8141-8157页

作者： Li, Xuehan Jing, Tao Li, Ruinian Wang, Xiaoxuan Yan, Yu Fan, Xin Huo, Yan Yu, Fei Richard Beijing Jiaotong University School of Electronic and Information Engineering National Engineering Research Center of Advanced Network Technologies Beijing100044 China Bowling Green State University Department of CS Bowling GreenOH43403 United States Beijing Forestry University School of Information Science and Technology Beijing100083 China Carleton University Department of Systems and Computer Engineering OttawaONK1S 5B6 Canada

As vehicular edge computing and networks (VECONs) emerge, Internet of vehicles (IoVs) is gaining significant advantages. To provide users with dependable services, VECONs need to offer efficient and secure task offloading services. Unfortunately, little effort has been put into addressing security issues such as the risk of privacy exposure when offloading data. Modeling and optimizing offloading dependability are affected by these issues. To address these concerns, a dependable three-layer offloading framework involving blockchain technology is presented in this paper, with fully homomorphic encryption (FHE) algorithm incorporated in order to mitigate privacy exposure risks. The framework divides tasks reasonably and introduces a novel measure of offloading dependability called cost from energy consumption, delay, and privacy exposure risk (cEDP) for constructing dependable offloading optimization. Additionally, a self-learning based dependable privacy preservation offloading (SDPPO) algorithm is presented that is capable of solving the optimal offloading problem in a decentralized manner. Simulation results show that the SDPPO algorithm outperforms its counterparts in terms of faster convergence speed, better cEDP convergence results, and improved scalability and tunability characteristics. © 1967-2012 IEEE.

关键词： Differential privacy

来源：评论

学校读者我要写书评

暂无评论

Complex Singular Spectrum Analysis Leveraging Adaptive Taper Windows for Enhancing Mode Reconstruction From Multivariate Signals

引用

IEEE Signal Processing Letters 2025年 32卷 1820-1824页

作者： Gu, Jialiang Hung, Kevin Ling, Bingo Wing-Kuen Chow, Daniel Hung-Kay Zhou, Yang Hong Kong Metropolitan University Department of Electronic Engineering and Computer Science School of Science and Technology Hong Kong Hong Kong Tsientang Institute for Advanced Study Center for the Integrated Circuits and Artificial Intelligence Hangzhou310024 China The Education University of Hong Kong Department of Health and Physical Education Faculty of Liberal Arts and Social Sciences Hong Kong Hong Kong Guangdong Ocean University School of Electronic and Information Engineering Zhanjiang524088 China

In this letter, a generic extension of complex singular spectrum analysis (CSSA), referred to as GC-SSA, is proposed to enhance mode reconstruction from multivariate signals. This is achieved by introducing adaptive taper windows for CSSA. Specifically, we formulate an optimization problem related to window design for specific multivariate signals, and then employ an iterative algorithm to optimize the coefficients of the taper windows. GC-SSA using optimized taper windows can decompose multivariate signals and perfectly reconstruct time-varying modes that have maximally concentrated energy. Numerical simulations were demonstrated to validate the effectiveness of the proposed method in mode reconstruction compared to other multivariate signal processing methods. © 1994-2012 IEEE.

关键词： Image reconstruction

来源：评论

学校读者我要写书评

暂无评论

Big Data Clustering Algorithms: Improving Efficiency and Scalability for Massive Data Mining and Pattern Recognition

Big Data Clustering Algorithms: Improving Efficiency and Sca...

引用

2025 International Conference on Pervasive Computational Technologies, ICPCT 2025

作者： Muthu, Shanmugam Kalyan, G Ramesh Samuthira Pandi, V. Shobana, D. Lakshmi Priya, J. Jaishika, S. Shipt Inc 420 20th St N # 100 BirminghamAL35203 United States Sns College of Technology Department of Computer Science and Engineering Tamilnadu Coimbatore India Chennai Institute of Technology Centre for Advanced Wireless Integrated Technology Tamil Nadu Chennai India Rajalakshmi Engineering College Department of Mechatronics Engineering Chennai India Saveetha Engineering College Department of Ece Chennai India Vel Tech Multi Tech Dr.Rangarajan Dr.Sakunthala Engineering College Department of Computer Science and Engineering India

ISBN: (纸本)9798331508685

The emergence of big data has completely changed the way data analysis is done. As a result, there is a pressing need for effective clustering algorithms that can handle large datasets from different fields. Efficiency, scalability, and finding meaningful patterns in enormous data are some of the important difficulties that this research tackles in relation to big data clustering. We thoroughly examine and evaluate a range of clustering algorithms, from more conventional ones like hierarchical and K-Means clustering to more cutting-edge ones like density-based spatial clustering (DBSCAN) and model-based clustering. To further improve the performance and scalability of clustering algorithms, we investigate new methods that make use of distributed computing frameworks like Apache Spark and MapReduce. We offer a thorough assessment of their efficacy in terms of clustering quality, computational efficiency, and resource use by using and evaluating these methods on large-scale datasets. While conventional clustering algorithms may fail miserably when faced with extremely big data sets, our research shows that newer methods that make use of optimization techniques and distributed processing greatly outperform their predecessors. The effect of feature selection and dimensionality reduction techniques on improving clustering results is also shown. By shedding light on how to choose and implement clustering algorithms for effective data mining and pattern identification, this study adds to the continuing endeavors in the domain of big data analytics. Progress in areas like customer segmentation, anomaly detection, and social network analysis can be achieved with the help of adaptive algorithms that can handle large datasets while keeping clustering results accurate and dependable, according to the study. © 2025 IEEE.

关键词： Data accuracy

来源：评论

学校读者我要写书评

暂无评论

RADAR: Robust DRL-Based Resource Allocation Against Adversarial Attacks in Intelligent O-RAN

IEEE Transactions on Green Communications and Networking

引用

IEEE Transactions on Green Communications and Networking 2025年

作者： Ergu, Yared Abera Nguyen, Van-Linh Department of Computer Science and Information Engineering Chiayi Taiwan Advanced Institute of Manufacturing with High-tech Innovations CCU Taiwan

The advent of open radio access networks (O-RAN) has introduced intelligent, flexible, and multi-vendor network ecosystems. While O-RAN's open interfaces and artificial intelligence (AI)-driven solutions offer improved performance, energy efficiency, and resource minimization for green networking, they also expose the system to new security vulnerabilities, particularly adversarial attacks. This paper presents a robust defense approach, termed RADAR, designed to secure deep reinforcement learning (DRL)-powered resource allocation mechanisms in O-RAN. RADAR is a multi-faceted defense framework that integrates adversarial input sanitization, proactive adversarial training, and adapted defensive distillation to counter policy infiltration attacks, gradient-based deceptive loss maximization, and signal perturbation injections into the O-CU via the O-DU in O-RAN. This study evaluates the effectiveness of RADAR not only against a novel attack variant - policy infiltration attack (PIA), which manipulates environmental parameters to disrupt allocation decisions, but also against well-known adversarial techniques such as the fast gradient sign method (FGSM) and projected gradient descent (PGD). Experimental results demonstrate that RADAR achieves significant recovery in user data rates across three network slices: 73.33% for eMBB, 64.71% for mMTC and 52.94% for uRLLC, outperforming the existing standalone approach. The findings highlight RADAR's effectiveness in mitigating adversarial attack techniques, underscoring its potential to secure AI-driven core functions in intelligent O-RAN. © 2025 IEEE.

关键词： Energy efficiency

来源：评论

学校读者我要写书评

暂无评论

Decentralized Authentication Scheme Incorporating Reputation and Attribute Signature for Cross-Web3 Applications

引用

IEEE Internet of Things Journal 2025年第10期12卷 14526-14537页

作者： Song, Xiaoling Xu, Guangxia Huang, Yongfei Liu, Yong Chongqing University of Posts and Telecommunications School of Computer Science and Technology Chongqing400065 China Guangzhou University Cyberspace Institute of Advanced Technology Guangzhou510006 China Peng Cheng Laboratory Department of New Networks Shenzhen518000 China Chongqing University of Posts and Telecommunications School of Software Engineering Chongqing400065 China

Driven by Web 3.0, the number of decentralized applications (DAPPs) continues to increase, and the need for decentralized authentication across applications becomes more and more significant. In addition, multi-dimensional reputation assessment becomes especially critical in a decentralized environment. However, there is still a lack of trusted and flexible self-sovereign identity (SSI) authentication mechanisms in Web 3.0 decentralized zero-trust cross-applications access scenarios. In response to these challenges, we propose a decentralized authentication scheme that fuses reputation assessment and attribute signature for the open and self-organized Web3 application. Firstly, we design an attribute signature method based on the attributes in a user's verifiable credentials (VCs). The signature can be used for authentication across Web3 applications, and the authenticator can flexibly adjust the attributes to be verified and their number according to specific needs. Meanwhile, in the attribute signature, we construct a multi-dimensional reputation assessment mechanism to comprehensively assess the user's reputation from multiple aspects to determine his/her security risk level, so as to reduce the possibility of identity fraud. Ultimately, through comparative analysis, we verify the advantages of the scheme in terms of privacy, scalability and attack resistance. Moreover, we analyze the performance of the key algorithms in the scheme through simulation experiments, and the experimental results show that our scheme is feasible. © 2014 IEEE.

关键词： Authentication Protocol

来源：评论

学校读者我要写书评

暂无评论

Convolutional Neural Network for Facial Emotion Detection

Convolutional Neural Network for Facial Emotion Detection

引用

IT Innovation and Knowledge Discovery (ITIKD), International Conference on

作者： Tina Babu Rekha R Nair A G Kavya Tenzin Sungrab Norbu Sunil Reddy J Harsha Vardhan Reddy Department of Computer Science and Engineering Alliance School of Advanced Computing Alliance University Bengaluru India

ISBN: (数字)9798350355468

ISBN: (纸本)9798350355475

Facial Emotion Recognition is considered one of the important domains in computer vision, which allows machines to interpret human emotions and, consequently, find applications in adaptive gaming, healthcare, and so forth. In this work, a Convolutional Neural Network (CNN)-based approach to emotion classification using FER-2013 dataset is proposed. This dataset contains 35,887 grayscale images with seven categories of emotions such as angry, disgust, fear, happy, sad, surprise, and also neutral. The architecture integrates convolutional layers, max-pooling, dropout regularization, and fully connected layers for robust feature extraction and classification. Data augmentation techniques in terms of rotation, zooming, and flipping are applied to reduce overfitting and enhance generalization. Training uses the Adam optimizer with a decaying learning rate, in addition to categorical cross-entropy loss, in a total number of 30 epochs. Experiments are performed to get a 54.7% accuracy on the validation set; strong performance is achieved for emotions “happy” and “surprise,” whereas “neutral” and “fear” cannot be well distinguished. The experiment stresses that one ought to balance datasets, exploit multimodal data and advanced model optimization techniques for improved accuracy and making models useful in the real world. Findings set the benchmark for upgrades in FER into better robust and scalable systems to cater to varying demands for applications.

关键词： Training Emotion recognition Data privacy Accuracy Computational modeling Medical services Data augmentation Data models Convolutional neural networks Optimization

来源：评论

学校读者我要写书评

暂无评论

Hybrid Approach: Combining Hill Climbing and Genetic Algorithms for Traveling Salesman Problem

Hybrid Approach: Combining Hill Climbing and Genetic Algorit...

引用

IT Innovation and Knowledge Discovery (ITIKD), International Conference on

作者： Tina Babu Rekha R Nair S Sumanth Ishith T P Chitra K S Department of Computer Science and Engineering Alliance School of Advanced Computing Alliance University Bengaluru India

ISBN: (数字)9798350355468

ISBN: (纸本)9798350355475

The Traveling Salesman Problem (TSP) is a classic NP-hard optimization challenge, requiring an efficient approach to identify near-optimal solutions. This paper presents a hybrid algorithm combining Genetic Algorithm (GA) and Hill Climbing (HC) to improve solution accuracy and convergence speed. GA provides global search capabilities through selection, crossover, and mutation, while HC enhances local optimization by refining solutions iteratively. The hybrid approach leverages GA's exploration strength and HC's exploitative power, reducing premature convergence and improving search efficiency. Experimental results indicate that the hybrid GA-HC algorithm outperforms standalone GA and HC in terms of tour length reduction and computational efficiency. Benchmarked on TSP instances, the proposed method demonstrates a 15% improvement in solution quality compared to traditional approaches. Additionally, the hybrid method achieves faster convergence, requiring 30% fewer iterations than standalone GA. The results highlight the hybrid algorithm's potential for real-world applications such as logistics, network design, and supply chain management. This study contributes to the growing research on hybrid metaheuristics, showcasing the benefits of integrating global and local search strategies to solve complex combinatorial problems.

关键词： Traveling salesman problems Benchmark testing Search problems Hybrid power systems Computational efficiency Optimization Tuning Genetic algorithms Convergence Logistics

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：