检索结果-内蒙古大学图书馆

MBPD: A Robust Algorithm for Polar-Domain Channel Estimation in Near-Field Wideband XL-MIMO Systems

IEEE Internet of Things Journal 2024年第12期12卷 18461-18470页

作者： Wang, Han Guo, Peiqing Li, Xingwang Wen, Fangqing Wang, Xianpeng Nallanathan, Arumugam Yichun University College of Physical Science and Engineering Yichun336000 China Nanjing University of Posts and Telecommunications School of Computer Science Nanjing210023 China Henan Polytechnic University School of Physical and Electronic Information Engieering Jiaozuo454000 China China Three Gorges University College of Computer and Information Technology Yichang443002 China Hainan University School of Information and Communication Engineering Haikou570228 China Queen Mary University of London School of Electronic Engineering and Computer Science LondonE1 4NS United Kingdom Kyung Hee University Department of Electronic Engineering Gyeonggi-do Yongin-si17104 Korea Republic of

In the evolving landscape of wireless communications, extremely large-scale multiple-input multiple-output (XL-MIMO) systems offer promising enhancements in capacity and spectral efficiency, particularly in near-field scenarios. This paper investigates polar-domain channel estimation methods for near-field wideband XL-MIMO systems, proposing a novel approach based on the bilinear pattern detection (BPD) method. We introduce the multi-candidate BPD (MBPD) algorithm, which improves detection accuracy by incorporating adaptive weight matrix adjustments and evaluating multiple candidate modes per iteration. Comprehensive simulations validate the superiority of MBPD over traditional BPD in terms of estimation accuracy and robustness. Furthermore, a detailed complexity analysis demonstrates the computational feasibility of the proposed algorithm. The MBPD algorithm greatly improves polar-domain channel estimation, facilitating more efficient implementations of near-field wideband XL-MIMO systems. © 2024 IEEE.

关键词： Channel estimation

来源：评论

学校读者我要写书评

暂无评论

Individualization Of Head Related Transfer Function Based On PCA And RBF Network 22

Individualization Of Head Related Transfer Function Based On...

引用

Proceedings of the 2022 11th International Conference on Networks, Communication and Computing

作者： Wei Chen Hongxu Zhang Jinxia Yu Fen Luo School of Software Henan Polytechnic University China School of Computer Science and Technology Henan Polytechnic University China

ISBN: (纸本)9781450398039

Head-Related Transfer Function (HRTF) describes the acoustic reflection and diffraction effect caused by the influence of the human body (head, torso, etc.) in the transmission of sound waves to the human ear. In Virtual Reality(VR) / Augmented Reality(AR), HRTF is often used to generate virtual 3D audio due to its ability to recreate perceptions of natural sound scenes realistically. However, HRTF varies from person to person due to the differences in anthropometric features. Using non-individualized HRTF to produce 3D sounds may lead to hearing localization bias in users. Therefore, how to obtain individualized HRTF is a hot topic in the field of VR / AR. This paper proposes an effective method to establish the relationship between anthropometric features and HRTF. At first, a method based on multimodal principal component analysis is proposed for the representation of HRTF models with low dimensions. Then a nonlinear mapping representation model between the low-dimensional features of HRTF and anthropometric features is established using Radial Basis Function Neural Network (RBFNN). Objective experiments show that the proposed HRTF Individualization method can reduce the spectral distortion as low as 4.48 dB. The subjective listening experiments based on the principal sagittal plane show that the individualized HRTF obtained using this method can effectively improve the accuracy of subjective listening (about 33%).

关键词： RBFNN Individualization HRTF PCA

来源：评论

学校读者我要写书评

暂无评论

Hearing loss classification via AlexNet and extreme learning machine

引用

International Journal of Cognitive Computing in Engineering 2021年 2卷 144-153页

作者： Li, Bin School of Computer Science and Technology Henan Polytechnic University Henan Jiaozuo 454000 China

Hearing is a very important feeling for humans. It helps to perceive the environment around us and can warn us of any imminent dangers around us. Hearing loss is an impaired ability to hear sound. Sensorineural hearing loss (SNHL) is very common in today's society. Classification of hearing loss is helpful for clinical diagnosis, finding suitable treatment methods and allocating appropriate medical interventions. This paper presents a novel method for hearing loss classification based on magnetic resonance images, which can automatically recognize tissue specific features in a given MRI image. The study involved magnetic resonance images (MRI) of 60 participants in three categories of balance: left SNHL, right SNHL, and a healthy control group. Our method uses AlexNet to extract the features of a single category, and uses Extreme Learning Machine to form a triclassifier for automatic SNHL classification. From the experimental results and the practicability of the algorithm, the classification performance is obviously better than the standard deep learning method and other traditional methods. © 2021

关键词： AlexNet Extreme learning machine Hearing loss classification SNHL

来源：评论

学校读者我要写书评

暂无评论

Low-light Image Enhancement via Unsupervised Progressive Fusion Network 22

Low-light Image Enhancement via Unsupervised Progressive Fus...

引用

Proceedings of the 2022 11th International Conference on Networks, Communication and Computing

作者： Hengxuan Li Zhanqiang Huo Yingxu Qiao School of Software Henan Polytechnic University China School of Computer Science and Technology Henan Polytechnic University China

ISBN: (纸本)9781450398039

To address the issue that unsupervised low-light image enhancement methods rely on finely selected training sets or contrast exposure stacks, this paper proposes a low-light image enhancement network with unsupervised progressive fusion driven jointly by knowledge and data. The network consists of low-light image pre-processing stage and progressive fusion stage. In the pre-processing stage, the input low-light image is first inverted. Next, we perform three methods of fixed-value gamma correction, adaptive gamma correction, and color-preserving adaptive histgram equalization (color-preserving AHE) to generate three images with different degrees of contrast enhancement. The progressive fusion stage performs a progressive fusion of the three images generated in the preprocessing stage and the inverted input image. The contrast and brightness of the fused result are improved by fusion while suppressing overexposure in the bright regions. It is shown experimentally that the proposed method outperforms the current state-of-the-art unsupervised low-light image enhancement networks in terms of both visual quality and technical specifications. In some low-light scenes, our method even outperforms the supervised learning-based low-light image enhancement network.

关键词： Unsupervised learning Low-light image enhancement Image fusion

来源：评论

学校读者我要写书评

暂无评论

BEVUDA++: Geometric-Aware Unsupervised Domain Adaptation for Multi-View 3D Object Detection

引用

IEEE Transactions on Circuits and Systems for Video technology 2025年第5期35卷 5109-5122页

作者： Zhang, Rongyu Liu, Jiaming Li, Xiaoqi Chi, Xiaowei Wang, Dan Du, Li Du, Yuan Zhang, Shanghang Nanjing University School of Electronic Science and Engineering Nanjing210093 China The Hong Kong Polytechnic University Department of Computing Hung Hom Hong Kong Peking University National Key Laboratory for Multimedia Information Processing School of Computer Science Beijing100871 China The Hong Kong University of Science and Technology Academy of Interdisciplinary Studies Hung Hom Hong Kong

Vision-centric Bird’s Eye View (BEV) perception holds considerable promise for autonomous driving. Recent studies have prioritized efficiency or accuracy enhancements, yet the issue of domain shift has been overlooked, leading to substantial performance degradation upon transfer. We identify major domain gaps in real-world cross-domain scenarios and initiate the first effort to address the Domain Adaptation (DA) challenge in multi-view 3D object detection for BEV perception. Given the complexity of BEV perception approaches with their multiple components, domain shift accumulation across multi-geometric spaces (e.g., 2D, 3D Voxel, BEV) poses a significant challenge for BEV domain adaptation. In this paper, we introduce an innovative geometric-aware teacher-student framework, BEVUDA++, to diminish this issue, comprising a Reliable Depth Teacher (RDT) and a Geometric Consistent Student (GCS) model. Specifically, RDT effectively blends target LiDAR with dependable depth predictions to generate depth-aware information based on uncertainty estimation, enhancing the extraction of Voxel and BEV features that are essential for understanding the target domain. To collaboratively reduce the domain shift, GCS maps features from multiple spaces into a unified geometric embedding space, thereby narrowing the gap in data distribution between the two domains. Additionally, we introduce a novel Uncertainty-guided Exponential Moving Average (UEMA) to further reduce error accumulation due to domain shifts informed by previously obtained uncertainty guidance. To demonstrate the superiority of our proposed method, we execute comprehensive experiments in four cross-domain scenarios, securing state-of-the-art performance in BEV 3D object detection tasks, e.g., 12.9% NDS and 9.5% mAP enhancement on Day-Night adaptation. © 1991-2012 IEEE.

关键词： Teaching

来源：评论

学校读者我要写书评

暂无评论

Hand Multimodal Recognition Based on Adaptive Multichannel Graph Convolutional Networks 5

Hand Multimodal Recognition Based on Adaptive Multichannel G...

引用

5th IEEE International Conference on Pattern Recognition and Machine Learning, PRML 2024

作者： Lai, Xinbo Abdireyim, Alimjan Kasenmu, Reyihanguli Yadikar, Nurbiya Ubul, Kurban School of Computer Science and Technology Xinjiang University Urumqi China Science and Technology Project Service Centre Xinjiang Uighur Autonomous Region Urumqi China School of Information Science and Technology Xinjiang Normal and Higher Polytechnic University Urumqi China

ISBN: (纸本)9798350355925

In the biometrics field, the unimodal biometrics recognition effect cannot meet the requirements for high-performance identity recognition due to its characteristics, such as instability and limitations. Multimodal biometrics combines two or more biometric features to improve the accuracy and security of identification, and combining multiple biometric features can make up for the shortcomings of unimodal biometrics. Aiming at the above problems, this paper proposes a hand multimodal recognition network based on an adaptive multichannel graph convolutional network, which adopts the PageRank centrality fusion method for image fusion of fingerprints, finger veins, and palm veins so that the network takes into account both the feature novelty information and the structural information in the image fusion process. At the same time, we introduce the CBAM attention module in the network to allow the network to adaptively learn important features in channel and spatial dimensions to improve the network's performance. The experimental results show that the recognition accuracy of the method proposed in this paper reaches 99.96%, which has high performance and can be widely used in practical applications. © 2024 IEEE.

关键词： Image fusion

来源：评论

学校读者我要写书评

暂无评论

MDSTransUNet: Multi-Scale Deep Supervised Transformer U-Net for COVID-19 Lung and Infection Segmentation

SSRN

引用

SSRN 2025年

作者： Yan, Yidan Hou, Beibei Sun, Junding School of Computer Science and Technology Henan Polytechnic University No.2001 Century Road Henan Jiaozuo454000 China

The global COVID-19 pandemic outbreak in early 2020 precipitated an unprecedented public health crisis, highlighting the urgent need for advanced diagnostic solutions. Hybrid architecture based on U-Net and transformer has attracted widespread attention in medical image segmentation. However, the global attention mechanism compromise local edge information and tend to overlook subtle lung infection regions. In this paper, we propose a novel multi-scale deep supervised segmentation network, namely MDSTransUNet, which is designed for segmenting lung and infection areas from chest CT and X-ray images. Specifically, MDSTransUNet follows the traditional encoding-decoding paradigm, integrating ShuffleNet in conjunction with TransFormer as the encoder. It also utilizes inputs of diverse image sizes for deep feature supervision. Two decoder configurations are established to segment lung tissue and infected regions simultaneously. Furthermore, three pivotal modules are proposed: the multi-scale parallel pooling module (MSPP), the enhanced atrous spatial pyramid pooling module (EASPP), and the global context channel attention module (GCCAM). These modules serve to reinforce feature learning, extract multi-scale features, and acquire global attention, thereby enhancing the segmentation accuracy. A substantial number of experiments across two modalities images demonstrate that our model segment lung regions of varying sizes and pneumonia infection areas precisely, exhibiting high accuracy and outperforming other comparative methods. © 2025, The Authors. All rights reserved.

关键词： COVID-19

来源：评论

学校读者我要写书评

暂无评论

Convolutional neural network based data interpretable framework for Alzheimer’s treatment planning

引用

Visual Computing for Industry,Biomedicine,and Art 2024年第1期7卷 375-386页

作者： Sazia Parvin Sonia Farhana Nimmy Md Sarwar Kamal Information Technology Melbourne PolytechnicMelbourne VIC 3072Australia Faculty of Economics and Business University of New South WalesSydney ACT 2612Australia School of Computer Science Faculty of Engineering and ITUniversity of Technology SydneySydney NSW 2007Australia

Alzheimer’s disease(AD)is a neurological disorder that predominantly affects the *** the coming years,it is expected to spread rapidly,with limited progress in diagnostic *** machine learning(ML)and artificial intelligence(AI)algorithms have been employed to detect AD using single-modality ***,recent developments in ML have enabled the application of these methods to multiple data sources and input modalities for AD *** this study,we developed a framework that utilizes multimodal data(tabular data,magnetic resonance imaging(MRI)images,and genetic information)to classify *** part of the pre-processing phase,we generated a knowledge graph from the tabular data and MRI *** employed graph neural networks for knowledge graph creation,and region-based convolutional neural network approach for image-to-knowledge graph ***,we integrated various explainable AI(XAI)techniques to interpret and elucidate the prediction outcomes derived from multimodal ***-wise relevance propagation was used to explain the layer-wise outcomes in the MRI *** also incorporated submodular pick local interpretable model-agnostic explanations to interpret the decision-making process based on the tabular data *** expression values play a crucial role in AD *** used a graphical gene tree to identify genes associated with the ***,a dashboard was designed to display XAI outcomes,enabling experts and medical professionals to easily comprehend the predic-tion results.

关键词： Multimodal Region-based convolutional neural network Layer-wise relevance propagation Submodular pick local interpretable model-agnostic explanations Graphical genes tree Alzheimer’s disease

来源：评论

学校读者我要写书评

暂无评论

Recording Behaviors of Artificial Intelligence in Blockchains

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2023年第6期4卷 1437-1448页

作者： Zhang, Yushu Zhao, Jiahao Jiang, Jiajia Zhu, Youwen Wang, Liangmin Xiang, Yong Nanjing University of Aeronautics and Astronautics College of Computer Science and Technology Nanjing210023 China Henan Key Laboratory of Network Cryptography Technology Zhengzhou450001 China Southeast University School of Cyber Science and Engineering Nanjing211189 China Deakin University School of Information Technology MelbourneVIC3125 Australia

The key challenges of current blockchain, such as low throughput, high latency, and poor scalability, hinder its future development. The recent advances in artificial intelligence (AI) can well overcome these challenges, thus spurring an increasing number of organizations and individuals to equip blockchain with AI, making blockchain become more intelligent. However, AI behaviors need to be supervised for ex-post forensics in case of dispute and accountability. This motivates us to envision a model to record the AI behaviors in intelligent blockchain. In this article, we propose such a model called AI-Tracer, which leverages proxy re-encryption based on Schnorr signature as well as InterPlanetary File System (IPFS) to be integrated into intelligent blockchain. AI-Tracer generates AI-digest during the AI learning process and encrypts it with proxy re-encryption. IPFS is responsible to store the encrypted AI-digest and returns a hash pointer related to AI-digest for further verification. The authorized user can verify the validity of AI-digest by leveraging advanced features of blockchain technology and only the valid AI-digest can be uploaded to intelligent blockchain, which implements trusted tracking for AI behaviors. AI-Tracer achieves fine-grained access control of AI-digest via proxy re-encryption. A user can retrieve the plaintext of AI-digest from intelligent blockchain with specific authorization. We conduct theoretical analysis to indicate the privacy and security of AI-Tracer. Moreover, we deploy AI-Tracer on the Hyperledger Fabric and demonstrate the efficiency and effectiveness of AI-Tracer through extensive experiments. © 2020 IEEE.

关键词： Smart contract

来源：评论

学校读者我要写书评

暂无评论

Systematic framework for short text classification based on improved TWE and supervised MCFS topic merging strategy

引用

International Journal of computers and Applications 2022年第5期44卷 401-413页

作者： Sun, Baoshan Ge, Mengying Zhao, Peng Li, Chunqing Tianjin Key Laboratory of Autonomous Intelligence Technology and Systems School of computer science and technology Tianjin Polytechnic University Tianjin China

Text classification task can help people to discover valuable information hidden in the text set. Many previous studies have achieved excellent results in traditional text classification tasks. With the development of new social media, a large number of short texts appear on the Internet. Due to the sparsity of the short text, many classification algorithms which achieve excellent results on long texts hardly achieve satisfactory results on short texts. Therefore, it is important to find a method for calculating effective vector representations of words and overcoming the feature sparseness problem. Based on the above, we carry out the work from improving the quality of word vector representation and enhancing the effect of classification. This paper proposes a systematic framework for improving short text classification performance. In our framework, we first build a topic model with Latent Dirichlet Allocation (LDA) on a universal dataset from Wikipedia and use this model to perform topic inference on short texts. Then, an improved scheme of topical word embedding (TWE) is proposed to learn the vector representations of both words and topics, which use the word in the current word-topic pair to predict the contextual words and the topic in the same word-topic pair to predict its contextual topics. In addition, the supervised Multi-Cluster Feature Selection algorithm (MCFS) is employed to execute topic selection, and we propose a topic merging strategy that is based on the MCFS. At the end of topic selection and merging, short text matrices are generated using the vector representations of both words and topics, and these matrices are fed into a convolution neural network (CNN). On an open short text classification dataset, we compared the proposed framework with various baselines, and the experimental results indicate the effectiveness of our method. © 2020 Informa UK Limited, trading as Taylor & Francis Group.

关键词： Merging

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：