检索结果-内蒙古大学图书馆

2025 IEEE International Conference on Acoustics, Speech, and Signal processing, ICASSP 2025

作者： Li, Guohao Yang, Hongyu Men, Yifang Huang, Di Li, Weixin Yang, Ruijie Wang, Yunhong School of Computer Science and Engineering Beihang University Beijing China School of Artificial Intelligence Beihang University Beijing China Shanghai Artificial Intelligence Laboratory Shanghai China

ISBN: (纸本)9798350368741

Generating animatable and editable 3D head avatars is essential for various applications in computer vision and graphics. Traditional 3D-aware generative adversarial networks (GANs), often using implicit fields like neural Radiance Fields (NeRF), achieve photorealistic and view-consistent 3D head synthesis. However, these methods face limitations in deformation flexibility and editability, hindering the creation of lifelike and easily modifiable 3D heads. We propose a novel approach that enhances the editability and animation control of 3D head avatars by incorporating 3D Gaussian Splatting (3DGS) as an explicit 3D representation. This method enables easier illumination control and improved editability. Central to our approach is the Editable Gaussian Head (EG-Head) model, which combines a 3D Morphable Model (3DMM) with texture maps, allowing precise expression control and flexible texture editing for accurate animation while preserving identity. To capture complex non-facial geometries like hair, we use an auxiliary set of 3DGS and tri-plane features. Extensive experiments demonstrate that our approach delivers high-quality 3D-aware synthesis with state-of-the-art controllability. Our code and models are available at https://***/liguohao96/EGG3D. © 2025 IEEE.

关键词： 3D-aware GAN image video synthesis

来源：评论

学校读者我要写书评

暂无评论

A Real-Time Global Inference Network for One-Stage Referring Expression Comprehension

引用

IEEE Transactions on neural networks and Learning Systems 2023年第1期34卷 134-143页

作者： Zhou, Yiyi Ji, Rongrong Luo, Gen Sun, Xiaoshuai Su, Jinsong Ding, Xinghao Lin, Chia-Wen Tian, Qi Xiamen University Media Analytics and Computing Laboratory Xiamen361005 China Xiamen University School of Informatics Informatics and Communication Engineering Department Xiamen361005 China School of Informatics Xiamen University Media Analytics and Computing Laboratory Department of Artificial Intelligence Xiamen361005 China Xiamen University School of Informatics Xiamen361005 China National Tsing Hua University Department of Electrical Engineering Hsinchu30013 Taiwan Institute of Communications Engineering National Tsing Hua University Hsinchu30013 Taiwan Huawei Technologies Company Ltd. Cloud BU Shenzhen518129 China

Referring expression comprehension (REC) is an emerging research topic in computer vision, which refers to the detection of a target region in an image given a test description. Most existing REC methods follow a multistage pipeline, which is computationally expensive and greatly limits the applications of REC. In this article, we propose a one-stage model toward real-time REC, termed real-time global inference network (RealGIN). RealGIN addresses the issues of expression diversity and complexity of REC with two innovative designs: adaptive feature selection (AFS) and Global Attentive ReAsoNing (GARAN). Expression diversity concerns varying expression content, which includes information such as colors, attributes, locations, and fine-grained categories. To address this issue, AFS adaptively fuses features of different semantic levels to tackle the changes in expression content. In contrast, expression complexity concerns the complex relational conditions in expressions that are used to identify the referent. To this end, GARAN uses the textual feature as a pivot to collect expression-aware visual information from all regions and then diffuses this information back to each region, which provides sufficient context for modeling the relational conditions in expressions. On five benchmark datasets, i.e., RefCOCO, RefCOCO+, RefCOCOg, ReferIT, and Flickr30k, the proposed RealGIN outperforms most existing methods and achieves very competitive performances against the most advanced one, i.e., MAttNet. More importantly, under the same hardware, RealGIN can boost the processing speed by 10-20 times over the existing methods. © 2012 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

A Survey on Capsule networks: Evolution, Application, and Future Development

A Survey on Capsule Networks: Evolution, Application, and Fu...

引用

International Conference on High Performance Big Data and Intelligent Systems (HPBD and IS)

作者： Li, Jiawei Zhao, Qichen Li, Nan Ma, Lin Xia, Xuan Zhang, Xiaoguang Ding, Ning Li, Nannan Xi An Jiao Tong Univ Coll Comp Sci & Technol Xian Peoples R China Chinese Univ Hong Kong Sch Data Sci Shenzhen Peoples R China Shenzhen Inst Artificial Intelligence & Robot Soc Shenzhen Peoples R China Chinese Univ Hong Kong Inst Robot & Intelligent Mfg Shenzhen Peoples R China Macau Univ Sci & Technol Int Inst Next Generat Internet Macau Peoples R China

ISBN: (纸本)9781665413275

Capsule networks have drawn increasing attention since the concept is proposed in 2011. They hold advantages over convolutional neural networks (CNNs) on modeling part-to-whole relationships between entities and learning viewpoint invariant representations. A variety of improved designs and applications related to capsule networks have been explored within the past decade. This paper presents a relatively thorough survey of literatures corresponding to the evolution of the capsule networks and their applications. Different design of capsule networks from Hinton's group as well as other researchers are reviewed. applications of capsule networks in various scenarios and tasks are elaborated, including hyperspectral image processing, medical image processing, facial image processing, text classification, object detection, image segmentation, few-shot learning, etc. The relation and comparison of capsule networks, CNNs, and transformers, and the trend of future development of capsule networks are discussed. We believe this review provides a good reference to researchers who are interested in capsule networks.

关键词： Capsule networks deep learning artificial intelligence artificial neural network

来源：评论

学校读者我要写书评

暂无评论

Machine learning-based radar waveform classification for cognitive EW

引用

SIGNAL image AND VIDEO processing 2021年第8期15卷 1653-1662页

作者： Orduyilmaz, Adnan Yar, Ersin Kocamis, Mehmet Burak Serin, Mahmut Efe, Murat TUBITAK BILGEM ILTAREN TR-06800 Ankara Turkey Ankara Univ Dept Elect & Elect Engn TR-06560 Ankara Turkey

In this paper, we propose a waveform classification approach for cognitive electronic warfare applications in which a supervised classification method is presented in an efficient framework. In this manner, we introduce an end-to-end framework for detection and classification of radar pulses. Our approach is complete, i.e., we provide raw radar signal at the input side and produce categorical output in the end. We use short-time Fourier transform to obtain time-frequency image (TFI) of the signal. Hough transform is used to detect pulses in TFIs. Convolutional neural networks (CNN) are used for intra-pulse modulation classification. In experiments, we provide supervised classification results at different signal-to-noise ratio (SNR) levels and achieve 98.08% classification accuracy for 10 dB SNR on a diverse set of both frequency- and phase-modulated signals. The method sustains high classification accuracy levels as [93.9%;85.83%] for 0 and -10 dB SNR, respectively, that signifies the robustness of the method against noise for low-power intercepted signals. Simulation results also show that CNN outperforms artificial neural networks in intra-pulse modulation classification task.

关键词： Cognitive EW Supervised classification Intra-pulse modulation classification Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Power of image-Based Digit Recognition with Machine Learning 3rd

Power of Image-Based Digit Recognition with Machine Learning

引用

3rd International Conference on Emergent Converging Technologies and Biomedical Systems, ETBS 2023

作者： Abrol, Vipasha Nitika Pathak, Hari Gobind Shukla, Aditya Chandigarh University Gharuan India

ISBN: (纸本)9789819986453

Digit character recognition is recognizing and understanding written characters in images or scanned documents. This work often uses machine learning techniques such as SVM, KNN, and SNN to gain insights into large datasets of written examples of written characters. The purpose of the recognition process is to accurately convert images of written characters into digital representations that can be used for further processing or analysis. This technology is widely used in many applications such as OCR, signature recognition, and handwriting recognition. The main challenge in character recognition is developing an algorithm that can distinguish correct alphabets while resisting different patterns, sizes, and guidance. This is a challenge because the handwriting is different and there is often noise and distortion. The algorithm must be able to handle different types of input, such as scanned images, tablet typing, and camera images. One way to solve this problem is to use machine learning techniques such as Support Vector Machines (SVM), KNN, and artificial neural networks (ANN). The main purpose of this research is to provide an efficient and effective way to write code numbers. It is worth noting that even today's models and ideas are powerful, the problem of writing is important in the field of research, and new ideas and methods are constantly being produced. The results in this article show that the K-Nearest Neighbor (KNN) achieves the highest accuracy of 97.1%. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Model reduction in acoustic inversion by artificial neural network

引用

JOURNAL OF THE ACOUSTICAL SOCIETY OF AMERICA 2021年第5期150卷 3435-3444页

作者： Koponen, Janne Lahivaara, Timo Kaipio, Jari Vauhkonen, Marko Univ Eastern Finland Dept Appl Phys Kuopio Finland Univ Auckland Dept Math Auckland New Zealand Savonia Univ Appl Sci Sch Engn & Technol Kuopio Finland

In ultrasound tomography, the speed of sound inside an object is estimated based on acoustic measurements carried out by sensors surrounding the object. An accurate forward model is a prominent factor for high-quality image reconstruction, but it can make computations far too time-consuming in many applications. Using approximate forward models, it is possible to speed up the computations, but the quality of the reconstruction may have to be compromised. In this paper, a neural network-based approach is proposed that can compensate for modelling errors caused by the approximate forward models. The approach is tested with various different imaging scenarios in a simulated two-dimensional domain. The results show that with fairly small training datasets, the proposed approach can be utilized to approximate the modelling errors, and to significantly improve the image reconstruction quality in ultrasound tomography, compared to commonly used inversion algorithms.

关键词： Topics: Acoustics Descriptive statistics artificial neural networks Partial differential equations Speed of sound Tomography Numerical methods Signal processing Optical imaging

来源：评论

学校读者我要写书评

暂无评论

Design and Implementation of a Unified Power Data Measurement and Acquisition System Based on Deep neural Network

Design and Implementation of a Unified Power Data Measuremen...

引用

image processing, Electronics and Computers (IPEC), Asia-Pacific Conference on

作者： Yong Ma Ming Zhou Jing Zhang Zijian Zhang Jun Wang Yuyang Zhao Hang Cheng Information and Communication Branch State Grid Anhui Electric Power Company Hefei China

ISBN: (数字)9798350374407

ISBN: (纸本)9798350374414

This article introduces the design and implementation of a unified data measurement and acquisition system based on deep neural networks. With the advent of the big data era, data collection and processing have become increasingly important. However, existing data acquisition systems often face challenges due to the diversity of data sources and the complexity of data formats. To address these issues, we propose a unified data measurement and acquisition system based on deep neural networks. Firstly, we design the overall architecture of the system. The system consists of three main components: data acquisition module, data processing module, and data storage module. The data acquisition module is responsible for collecting data from various sources, including sensors, databases, and network interfaces. The data processing module uses deep neural networks to preprocess and extract features from the collected data. Finally, the data storage module saves the processed data to a database for subsequent analysis and applications. Secondly, we provide a detailed description of the system implementation process. We use a convolutional neural network-based data processing module to handle image data, and a recurrent neural network-based data processing module to handle time series data. By training and optimizing these modules, we are able to improve the accuracy and efficiency of the system. Lastly, we evaluate the system through experiments. We use a set of real-world datasets to test the performance of the system. The experimental results show that our system has high accuracy and efficiency in data collection and processing. Furthermore, we compare our system with other related systems, and the results demonstrate significant advantages in terms of performance. In conclusion, this article proposes a unified data measurement and acquisition system based on deep neural networks, which effectively addresses some challenges in data collection and processing. Through experi

关键词： Training Accuracy Databases Soft sensors Data acquisition Time series analysis Memory artificial neural networks Feature extraction System implementation

来源：评论

学校读者我要写书评

暂无评论

Design of Adaptive Control System Driven by artificial Intelligence and Its Application in Industry 4.0

Design of Adaptive Control System Driven by Artificial Intel...

引用

image processing and Computer applications (ICIPCA), IEEE International Conference on

作者： Xiaoqian Wen Business School Belarusian State University Minsk Belarus

ISBN: (数字)9798350360240

ISBN: (纸本)9798350384161

In view of the difficulty of effective fusion of multi-source sensing data used in industrial production, a method based on multi-source sensing data fusion was proposed. A multi-sensor network model for real-time monitoring in industrial field is established. The key techniques of multi-dimensional data extraction, self-fuzzy inference, multi-dimensional group estimation and multi-dimensional grouping mechanism are studied by four-level data analysis method. This improves the control accuracy and effectiveness of the multidimensional system. This scheme is based on I2C bus structure and uses single chip microcomputer to control multiple sensor nodes, which improves the software reconfiguration ability and robustness, and reduces the maintenance and updating cost of the system. This project intends to establish an adaptive control model by combining FLMS and BP neural network. According to the system error, the weight coefficients of the two kinds of neural networks are determined. In the initial stage, the convergence rate of the neural networks is improved first, and then the nonlinear BP network is used as the main assistant to improve the control effect.

关键词： Employee welfare Costs neural networks Microcomputers Hardware Real-time systems Software Sensors Time factors Adaptive control

来源：评论

学校读者我要写书评

暂无评论

Editorial: Deep learning in computational materials science

引用

FRONTIERS IN MATERIALS 2023年 10卷

作者： Xiao, Shaoping Bordas, Stephane Pierre Alain Kim, Tae-Yeon Univ Iowa Dept Mech Engn Iowa City IA 52242 USA Univ Luxembourg Doctoral Program Data & Computat Sci Luxembourg Luxembourg Khalifa Univ Sci & Technol Civil Infrastructure & Environm Engn Abu Dhabi U Arab Emirates

Riding the current wave of artificial intelligence (AI), many engineers and scientists have adopted machine learning (ML) and deep learning (DL) as powerful tools in various science and engineering disciplines, including materials science. Utilizing artificial neural networks (NNs), DL is one of the most effective, supervised, time-efficient, and cost-efficient ML approaches. It has been successfully used in various computational materials science research topics, such as developing potential functions for molecular dynamics, predicting the mechanics of materials, optimizing material and structural design, and promoting novel multiscale modeling and simulation. Moreover, DL enhances the data-driven approach in the materials science research community via big data analyses and image processing. This special issue aims to unite researchers to share insights into AI in current research projects and promote DL applications in computational materials *** learning uses statistical models to analyze data and draw inferences from its pattern. Particularly, supervised learning models learn the relationship between the input features and the output targets without explicit instructions. As a subset of ML, DL employs NNs to find appropriate representations from data for progressively good performance. In this special issue, Deshpande et al. proposed three types of NN architectures to learn nonlinear deformation efficiently to accelerate simulations in solid mechanics. The considered NN frameworks were based on convolutional neural networks (CNNs), multichannel aggregation networks (MAgNET), and attention-based neural networks, respectively. In addition, they employed two benchmark examples of largely deformed soft bodies and demonstrated the capabilities of the proposed methods to replace high-fidelity computational models in computational materials science. Since CNNs have been successfully utilized in image processing, they were also recently employed in computat

关键词： deep learning convolutional neural networks generative adversarial network materials science modeling and simulation

来源：评论

学校读者我要写书评

暂无评论

Ultrahigh bandwidth applications of optical microcombs 4

Ultrahigh bandwidth applications of optical microcombs

引用

AI and Optical Data Sciences IV 2023

作者： Tan, M. Sun, Y. Wu, J. Xu, X. Li, Yang Corcoran, B. Chu, S. Little, B. Morandotti, R. Mitchell, A. Moss, D.J. School of Engineering RMIT University MelbourneVIC3001 Australia Optical Sciences Centre Swinburne University of Technology HawthornVIC3122 Australia State Key Laboratory of Information Photonics and Optical Communications Beijing University of Posts and Telecommunications Beijing100876 China Department of Electrical and Computer Systems Engineering Monash University ClaytonVIC3800 Australia Department of Physics City University of Hong Kong Tat Chee Avenue Hong Kong Xi'an Institute of Optics and Precision Mechanics Precision Mechanics of CAS Xi'an China INRS-Énergie Matériaux et Télécommunications 1650 Boulevard Lionel-Boulet VarennesQCJ3X 1S2 Canada

ISBN: (纸本)9781510659810

We report ultrahigh bandwidth applications of Kerr microcombs at data rates beyond 10 Terabits/s. Optical neural networks can dramatically accelerate the computing speed to overcome the inherent bandwidth bottleneck of electronics. At the same time, digital signal processing has become central to many fields, from coherent optical telecommunications where it is used to compensate signal impairments, to image processing, important for observational astronomy, medical diagnosis, autonomous driving, big data and particularly artificial intelligence. Digital signal processing had traditionally been performed electronically, but new applications, particularly those involving real time video image processing, are creating unprecedented demand for ultrahigh performance, including bandwidth and reduced energy consumption. We use a new and powerful class of micro-comb called soliton crystals that exhibit robust operation and stable generation as well as a high intrinsic efficiency with a low spacing of 48.9 GHz. We demonstrate a universal optical vector convolutional accelerator operating at 11 Tera-OPS/s (TOPS) on 250,000 pixel images for 10 kernels simultaneously - enough for facial image recognition. We use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images. Finally, we demonstrate a photonic digital signal processor operating at 18 Tb/s and use it to process multiple simultaneous video signals in real-time. The system processes 400,000 video signals concurrently, performing 34 functions simultaneously that are key to object edge detection, edge enhancement and motion blur. As compared with spatial-light devices used for image processing, our system is not only ultra-high speed but highly reconfigurable and programable, able to perform many different functions without any change to the physical hardware. Our approach, based on an integrated Kerr solit

关键词： Digital signal processing

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：