检索结果-内蒙古大学图书馆

Disruptive Technologies (ICDT), International conference on

作者： Himanshu Sharma Devanand Padha Yashwant Singh School of CSET Bennett University Greater Noida India Department of CSE Model Institute of Engineering and Technology Jammu India Department of CSIT Central University of Jammu Jammu India

ISBN: (数字)9798331519582

ISBN: (纸本)9798331519599

Image captioning is an approach for precisely summarizing image contents using natural language processing. Leveraging cutting-edge feature extraction, multimodal encoding, and fusion strategies, visual and textual fragments are correlated within a latent space. These multimodal segments are analyzed by a deep sequential model for phrase decoding. The widely used CNN-LSTM encoder-decoder pipeline is central to image captioning frameworks. However, current architectures suffer from two major limitations: (1) existing frameworks employ a unidirectional fusion and sequential decoding process, failing to capture the semantic context against both the preceding and following tokens; and (2) static multimodal translation fails to grasp and align critical information accurately, limiting its effectiveness when training on diverse datasets with varied visual and linguistic samples. Consequently, the models fail to capture the expansive and intricate context, leading to the production of short and abstract captions. To address these limitations, we propose Bi-CANet, an end-to-end bidirectional multimodal contextual alignment network for image captioning. Bi-CANet employs a dual-directional fusion of visual details with textual phrases along with a bidirectional decoder, thereby aligning the rich semantic and spatial context over both the preceding and subsequent phrases. Experimental analysis shows that Bi-CANet is more effective in generating comprehensive captions when trained on diverse datasets. Specifically, it achieves BLEU-1 scores of 54.02%, and METEOR scores of 33.11% on the Flickr 8K dataset.

关键词： Training visualization Image databases Semantics Robustness Decoding Meteors Web sites Multimedia communication Context modeling

来源：评论

学校读者我要写书评

暂无评论

Enhanced Satellite Image Fusion Using Deep Learning and Feature Extraction Techniques: A Survey 1st

Enhanced Satellite Image Fusion Using Deep Learning and Feat...

引用

1st International conference on Intelligent Systems in Computing and communications, ISCComm 2023

作者： Nallagachu, Swathi Sandanalakshmi, R. Department of Electronics and Communication Engineering Puducherry Technological University Puducherry India

ISBN: (纸本)9783031756047

This paper presents an overview and analysis of numerous research projects on image fusion methods, with a particular emphasis on deep learning-based methods. The research analyses the inadequacies of current fusion models and suggests novel methods for feature extraction, visual sensor networks, remote sensing applications, medical imaging, and multi-resolution image fusion. The study demonstrates the advantages of deep learning techniques for image fusion tasks, including Convolutional Neural Networks (CNNs), Stacked Auto encoders, and Convolutional Sparse Representation (CSR). These methods provide better fusion quality, fast processing, and better visual perception. Numerous studies provide novel methods, such as the rapid Integer Lifting Wavelet Transform (ILWT), Dual-Tree Complex Contourlet Transform (DT-CCT), and hybrid fusion methods that combine Discrete Cosine Transform (DCT) and Integer Lifting Wavelet Transform (ILWT) techniques. The review literature survey shows how these techniques can improve fusion outcomes while introducing little information distortion. In general, this paper brings together improvements in feature extraction and image fusion techniques, demonstrating the strength and potential of deep learning-based approaches. It is a useful tool for academics and professionals working in various fields who want to comprehend and advance these fields of study. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Image fusion

来源：评论

学校读者我要写书评

暂无评论

DATA-VSR: Dynamic Trajectory Attention and Texture Adaptive Rooter for Video Super-Resolution

DATA-VSR: Dynamic Trajectory Attention and Texture Adaptive ...

引用

International conference on Acoustics, Speech, and Signal processing (ICASSP)

作者： Linfeng He Meiqin Liu Qi Tang Chao Yao Yao Zhao Institute of Information Science Beijing Jiaotong University Beijing China Visual Intelligence + X International Cooperation Joint Laboratory of MOE Beijing Jiaotong University Beijing China School of Computer and Communication Engineering University of Science and Technology Beijing Beijing China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Video Super-Resolution (VSR) is essential for reconstructing high-definition sequences from correlated video frames. While Transformer-based VSR methods have improved reconstruction quality, they require substantial computational resources, limiting deployment on resource-constrained devices. To tackle this issue, we propose a novel framework named Dynamic Trajectory Attention and Texture Adaptive Rooter for Video Super-Resolution (DATA-VSR). There are two key innovations: the Temporal Redundancy-aware Alignment Network (TRAN) and the Spatial Redundancy-aware Refinement Network (SRRN). Specifically, features are aligned by focusing on dynamic temporal trajectories instead of static redundancies in TRAN, and then features are adaptively refined based on the texture complexity of different regions in SRRN. Additionally, the Dual-Domain Enhancement Block (DDEB) is incorporated to effectively capture global dependencies in the frequency domain and enhance the representation of local features in the spatial domain. The experimental results on standard VSR benchmarks show that DATA-VSR achieves competitive performance with fewer parameters, lower FLOPs, and a specific reduction of 17%.

关键词： Technological innovation Limiting Superresolution Redundancy Transformers Trajectory Speech processing Spatial resolution Standards Signal resolution

来源：评论

学校读者我要写书评

暂无评论

Neural representation of allegedly sex-specific human body odor compounds

引用

NEUROIMAGE 2025年 310卷 121114-121114页

作者： Ferdenzi, Camille Fournel, Arnaud Fantin, Luca Ortegon, Stephane Richard Manesse, Cedric Baldovini, Nicolas Thevenet, Marc Lamberton, Franck Ibarrola, Danielle Faure, Frederic Bensafi, Moustafa Univ Claude Bernard Lyon 1 Ctr Rech Neurosci Lyon CRNL INSERM CNRSUMR5292U1028 Bron France Univ Cote Azur Inst Chim Nice CNRS UMR 7272 Nice France Univ Claude Bernard Lyon 1 CNRS INSERM SFR Lyon Est UAR3453US7 Lyon France CERMEP Imagerie Vivant Bron France Hop Edouard Herriot Hosp Civils Lyon Lyon France Infirm Protestante Caluire Et Cuire France

Body odors play an important role in nonverbal communication, and particularly in one's attractiveness. However, their central processing remains underexplored, especially as a function of gender. This study aims at identifying the neural networks involved in the processing of two allegedly sex-specific human body odor compounds (3-hydroxy-3-methylhexanoic acid, HMHA, and 3-methyl-3-sulfanylhexan-1-ol, MSH). We hypothesized that i) these body odors would be processed by different brain regions than non-body odors, and that ii) their role in attractiveness, if any, would be indicated by the activation of specific regions and by differential verbal and neurophysiological responses in men and women. Thirty participants (15 men, 15 women) performed a functional Magnetic Resonance Imaging (fMRI) session during which they rated the attractiveness of HMHA, MSH, and 5 non-body odorants. At the end of the session, participants rated all odors on multiple perceptual scales. HMHA activated visual (striate area and occipital gyrus) rather than olfactory brain regions. Men rated HMHA as more masculine than women did, and presented greater neural activity in the superior and medial frontal gyri while women activated the inferior frontal gyrus significantly more than men in response to this odor. MSH was processed as the other non-body odors, and not subject to gender differences. The results suggest that HMHA (not MSH) bears specific social information, resulting in a neural processing outside the main olfactory network. It is also processed differently in men and women, although our findings do not provide clear evidence in favor of relevance for one's attractiveness.

关键词： Human chemical communication Body odor Gender Attractiveness fMRI

来源：评论

学校读者我要写书评

暂无评论

3rd International conference on Electronic Governance with Emerging Technologies, EGETC 2024

3rd International Conference on Electronic Governance with E...

引用

3rd International conference on Electronic Governance with Emerging Technologies, EGETC 2024

ISBN: (纸本)9783031770289

The proceedings contain 17 papers. The special focus in this conference is on Electronic Governance with Emerging Technologies. The topics include: Ecological Development of a Novel Machine Learning Technique for Electricity Consumption and Cost Prediction in Cloud Computing;blockchain Based Decentralized Approach for Securing Fog Computing Environment;Enhancing Face Spoofing Detection Via CNN Model Integration with Normalized Features;Study of ZigBee-Based communication to Share Operative Weather information Between UASs in the Frame of Smart Cities Concept;automating Exam Assessment Through Digital Transformation and Image processing;an Analysis of Ethical Dimension of Machine Learning Models in Software Defect Prediction in Healthcare Sector;leaf Diseases Detection in Peach Using EfficientNet;medInsight: A Personalized Prescription Recommendation Using Machine Learning;intelligent Machine Learning for Cybersecurity: Anomaly Detection in Network Intrusion Systems and Beyond;Enhancing Cyber Resilience: Implementing DFMCAI for Threat Detection and Analysis;forecasting Future Credit Card Delinquencies: Leveraging Data Tables and visual Analysis for Enhanced Accuracy;navigating the Strategic Landscape: An In-Depth Review of the Strategic Planning Process;exploring Text Vectorization on the Polarity Detection of Spanish Comments;Evolutionary Optimization Techniques for the Design of Three-Dimensional UAV’s Antenna Arrays;cooperative Multi-agent Pursuit-Evasion Using Convolutional Neural Networks.

关键词：

来源：评论

学校读者我要写书评

暂无评论

ALIC: Adaptive Fusion Entropy Model for Learned Image Compression

ALIC: Adaptive Fusion Entropy Model for Learned Image Compre...

引用

International conference on Acoustics, Speech, and Signal processing (ICASSP)

作者： Lingxue Li Meiqin Liu Yifan Zhang Qi Tang Chao Yao Yao Zhao Institute of Information Science Beijing Jiaotong University Beijing China Visual Intelligence + X International Cooperation Joint Laboratory of MOE Beijing Jiaotong University School of Computer and Communication Engineering University of Science and Technology Beijing Beijing China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Recently, learned image compression algorithms have achieved significant performance. The entropy model is crucial for improving the rate-distortion performance by estimating the probability distribution of latent representation. In this paper, we propose an adaptive fusion entropy model for learned image compression (ALIC). To explore the correlation between channel and global spatial features, an adaptive fusion entropy model (AFEM) is designed. AFEM first slices the latent representation along the channels and leverages the adaptive channel fusion context module (ACFC) to capture correlations between the decoded and current slices. Subsequently, AFEM uses the adaptive spatial fusion context module (ASFC) to further divide the current slice into encoding pixels and reference pixels, thus improving the accuracy of probability estimation. The attention map and modulation parameter are introduced in ACFC and ASFC to interact with channel and spatial features. In addition, the variable-rate residual transformer (VResFormer) is proposed to control dynamic bit-rate by selectively modulating the high-frequency component according to coefficient weight and bias. Experimental results indicate that our ALIC outperforms other learned image compression algorithms. Our ALIC saves 5.89% bit-rate compared with VVC (4:4:4) on Kodak dataset.

关键词： Adaptation models Image coding Correlation Signal processing algorithms Rate-distortion Modulation Transformers Entropy Probability distribution Speech processing

来源：评论

学校读者我要写书评

暂无评论

GraphCom: communication Hierarchy-aware Graph Engine for Distributed Model Training 25

GraphCom: Communication Hierarchy-aware Graph Engine for Dis...

引用

Proceedings of the ACM on Web conference 2025

作者： Xinbiao Gan Tiejun Li Liang Wu Qiang Zhang Lingyun Song Bo Yang Jie Liu Kai Lu School of Computer National University of Defense Technology Changsha China National University of Defense Technology Changsha China School of Computer Science Northwest Polytechnical University Chanxi China

ISBN: (纸本)9798400712746

Efficient processing of large-scale graphs with billions to trillions of edges is essential for training graph-based large language models (LLMs) in web-scale systems. The increasing complexity and size of these models create significant communication challenges due to the extensive message exchanges required across distributed nodes. Current graph engines struggle to effectively scale across hundreds of computing nodes because they often overlook variations in communication costs within the interconnection hierarchy. This paper presents GraphCom, a communication-efficient message graph engine for graph processing on supercomputers. Our key idea is to leverage the network topology information to perform communication hierarchy-aware message aggregation, where messages are (i) gathered to the responsible nodes (referred to as monitors) in the source domains, (ii) transferred between monitors, and (iii) scattered to the target nodes in the target domains. GraphCom's aggregation is more aggressive in that each source domain (instead of the source node). We have implemented GraphCom on top of MPI. We demonstrate GraphCom's effectiveness with synthetic benchmarks and real-world graphs, utilizing up to 79,024 nodes and over 1.2 million processor cores, demonstrating that GraphCom surpasses leading graph- parallel systems and state-of-the-art counterparts in both throughput and scalability. Moreover, we have deployed GraphCom on a production supercomputer, where it consistently outperforms the top solutions on the Graph500 list. These results highlight the potential GraphCom has to significantly improve the efficiency of distributed large-scale graph-based LLM training by optimizing communication between distributed systems, making it an invaluable graph engine for distributed training tasks on web-scale graphs.

关键词： graph model

来源：评论

学校读者我要写书评

暂无评论

Deep Steganography Using SteganoGAN

Deep Steganography Using SteganoGAN

引用

Integrated Circuits and communication Systems (ICICACS), IEEE International conference on

作者： Devidas Bhat Suraj B Madagaonkar Vinayak G Pai Deepthi J. Shetty Information Science & Engineering NMAMIT Nitte Karnataka India Computer Science and Engineering IIT Dharwad Karnataka India Computer Science and Engineering Manipal Institute of Technology Manipal Karnataka India Department of CSE Nitte Meenakshi Institute of Technology Bangalore Karnataka India

ISBN: (数字)9798331508456

ISBN: (纸本)9798331508463

An improved method for covertly embedding secret messages within images is deep steganography utilizing SteganoGAN, which combines steganography and deep learning. SteganoGAN employs Generative Adversarial Networks (GANs) to create realistic images that can safely conceal secret data. The concealed message is integrated into the image pixels and encoded as a binary stream, which can be retrieved using a decoder, ensuring the final image maintains visual integrity. This technique provides strong security, as it is difficult to detect the hidden data, making it ideal for applications like digital water-marking, copyright protection, and private data transfer. The deep learning architecture enhances processing speed, increases data embedding capacity, and improves security. Experimental results using the Div2k dataset showed that models trained with a data depth of 4 achieved the highest bits per pixel (RS-BPP), demonstrating a balance between data embedding capacity and image quality (PSNR and SSIM). Despite its strengths, SteganoGAN faces challenges, such as vulnerability to attacks by adversaries familiar with the technique, emphasizing the need for continuous innovation to improve security.

关键词： Deep learning Training Steganography Ethics visualization Technological innovation Watermarking Generative adversarial networks Data models Security

来源：评论

学校读者我要写书评

暂无评论

Integrating Inception v3 and Retinex Enhancement for Accurate Diagnosis of Age-Related Macular Degeneration

Integrating Inception v3 and Retinex Enhancement for Accurat...

引用

International conference on Wireless communications, Signal processing and Networking (WiSPNET)

作者： Senthil K Sivagami S Vijaya Lakshmi R Sree Dedeepya B Department of Computer Science and Engineering Saveetha School of Engineering Saveetha UniversitySIMATS Chennai Tamil Nadu India Department of Information Technology Saveetha School of Engineering Saveetha UniversitySIMATS Chennai Tamil Nadu Department of AI&DS Saveetha School of Engineering Saveetha UniversitySIMATS Chennai Tamil Nadu

ISBN: (数字)9798331524227

ISBN: (纸本)9798331524234

Proactive identification of eye conditions remains essential to stop conditions from advancing and to enable proper medical treatment of retina-related disturbances. Over the recent past, scientists have dedicated significant attention to diagnosing eye diseases. visual impairments create enormous obstacles that reduce patients' overall life satisfaction because they affect their ability to complete daily duties. Age-related macular Degeneration (AMD) is one of the prevalent reasons behind permanent central vision loss in both eyes that affects people progressively without causing discomfort. The proposed model architecture applies pre-processing to fundus images, which are then integrated into a Retinex algorithm before combining them with InceptionV3 to obtain AMD diagnosis results. The algorithm manifests its capability to enhance contrast and noise reduction, and its procedures necessary imaging parameters for drusen diagnosis. The InceptionV3 architecture principles are integrated to boost the diagnostic accuracy of the research outcomes. The proposed model receives training through an Open Access Series of Images Reading database consisting of 5,000 images that contain various types of eye diseases. The accuracy achieved during 10-fold cross-validation measures 95.67%. The automated method for AMD detection requires consideration for its diagnostic value, which should help professionals initiate timely medical assistance.

关键词： Macular degeneration Deep learning Wireless communication Training Accuracy Signal processing Feature extraction Spatial resolution Medical diagnostic imaging Testing

来源：评论

学校读者我要写书评

暂无评论

Development of Web Platform for Home Screening of Neurological Disorders Using Artificial Intelligence

Development of Web Platform for Home Screening of Neurologic...

引用

Advances in Computer Science, Electrical, Electronics, and communication Technologies (CE2CT), International conference on

作者： S. Bama Praveen Kumar Maroju Subash Banala Sunil Kumar Sehrawat Mohanarajesh Kommineni Divya Kodi Department of CSE Velammal Engineering College Surapet Chennai Clientserver Technology Solutions Financial Services & Cloud Technologies Bausch Health Companies TEKsystems Global Services LLC Truist Financial

ISBN: (数字)9798331518578

ISBN: (纸本)9798331518585

Millions of people worldwide suffer with Parkinson's disease (PD), one of the most prevalent neurological illnesses. The diagnosis of PD has changed significantly over the last 25 years; at first, autopsy and neurological exams were the only methods accessible. Impaired handwriting has been shown to be a significant early indicator for this disease's identification. The purpose of this study was to use machine learning techniques to provide a quick and easy method of differentiating PD patients from controls using handwriting activities. It developed a telemonitoring technology based on user-friendly drawing tablet software that gave the subject visual feedback on the screen and enabled us to gather data in real time about the location, pressure, and tilt of the digital pen throughout the procedure. This paper created a procedure that included writing and drawing exercises, including ones in Italian, and we gathered information from ten PD patients and twenty healthy participants. With a 93.4% accuracy rate, the current research developed a machine learning (ML) framework that automatically separates PD patients from healthy control people using signals and data from a previous database.

关键词： Neurological diseases visualization Accuracy Databases Machine learning Writing Signal processing Particle measurements Software Real-time systems

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：