检索结果-内蒙古大学图书馆

IEEE Transactions on Audio, Speech and Language Processing 2025年 33卷 2125-2138页

作者： Zhengtao Yu Jinlei Xu Xiang Huang Zhongfen Deng Yonghua Wen Li Sun Yuxin Huang Hao Peng Philip S. Yu Faculty of Information Engineering and Automation Kunming University of Science and Technology Kunming China Yunnan Province Key Laboratory of Artificial Intelligence Kunming University of Science and Technology Kunming China School of Cyber Science and Technology Beihang University Beijing China Department of Computer Science University of Illinois Chicago IL USA School of Control and Computer Engineering North China Electric Power University Beijing China

Multilingual Neural Machine Translation (NMT) excels in sharing knowledge across languages and transferring insights from high-resource languages to improve performance in low-resource languages. However, its performance lags in specific domains such as legal and medical. Previous works have focused on adding language-specific and domain-specific adapters to achieve domain adaptation. Although effective, these adapter-based methods only use domain data to train additional parameters, limiting the performance of multilingual NMT. In this paper, we propose CDSTX, a novel approach that achieves robust multilingual domain adaptation without relying on adapters, solely leveraging multilingual models. Specifically, we utilize the pretrained model XLM-R and frozen embeddings to preserve its robust multilingual capabilities and design a two-stage training strategy for domain adaptation. This includes separate training for the decoder and fine-tuning of the entire model, ensuring that the model effectively acquires domain knowledge and correctly represents domain-specific text. Moreover, to better utilize the domain information conveyed implicitly by the training data, we devise special domain tokens at the beginning of the source and target sentences called Source&Target Domain Tags. In addition, back translation is employed to enhance the cross-lingual transfer ability of our approach. Our proposed method is evaluated on two datasets across three translation tasks: Single-domain Multilingual NMT, Multi-domain bilingual NMT, and Multilingual Multi-domain NMT. Notably, in the single-domain multilingual NMT task, CDSTX significantly enhances zero-shot domain translation performance, achieving improvements of up to +30 BLEU points through the utilization of back-translation techniques. Even in the bilingual multi-domain NMT task where specific domain data for the target translation direction is unavailable, our method consistently outperforms all SOTA methods. Moreover, in the

关键词： Training Adaptation models Translation Limiting Law Large language models Training data Neural machine translation Multilingual Speech processing

来源：评论

学校读者我要写书评

暂无评论

Carbon-Aware Computing in a Network of Data Centers: A Hierarchical Game-Theoretic Approach

arXiv

引用

arXiv 2024年

作者： Breukelman, Enno Hall, Sophie Belgioioso, Giuseppe Dörfler, Florian The KTH Royal Institute of Technology School of Electrical Engineering and Computer Science Division of Decision and Control Systems Malvinas väg 10 StockholmSE-100 44 Sweden The Automatic Control Laboratory Department of Electrical Engineering and Information Technology ETH Zurich Physikstrasse 3 Zürich8092 Switzerland

Over the past decade, the continuous surge in cloud computing demand has intensified data center workloads, leading to significant carbon emissions and driving the need for improving their efficiency and sustainability. This paper focuses on the optimal allocation problem of batch compute loads with temporal and spatial flexibility across a global network of data centers. We propose a bilevel game-theoretic solution approach that captures the inherent hierarchical relationship between supervisory control objectives, such as carbon reduction and peak shaving, and operational objectives, such as priority-aware scheduling. Numerical simulations with real carbon intensity data demonstrate that the proposed approach successfully reduces carbon emissions while simultaneously ensuring operational reliability and priority-aware scheduling. © 2024, CC BY-NC-ND.

关键词： Carbon

来源：评论

学校读者我要写书评

暂无评论

AI-aided power electronic converters automatic online real-time efficiency optimization method

引用

Fundamental Research 2025年第3期5卷 1111-1116页

作者： Tang, Yuanhong Cao, Di Xiao, Jian Jiang, Chenying Huang, Qi Li, Yunwei Chen, Zhe Blaabjerg, Frede Hu, Weihao Power System Wide-area Measurement and Control Sichuan Provincial Key Laboratory School of Mechanical and Electrical Engineering University of Electronic Science and Technology of China Chengdu 611731 China MORNSUN Guangzhou Science & Technology Co. Ltd. Guangzhou 510663 China School of Information and Communication Engineering University of Electronic Science and Technology of China Chengdu 611731 China Yangtze Delta Region Institute (Quzhou) University of Electronic Science and Technology of China Quzhou 324000 China Department of Chengdu State Grid Corporation of China Chengdu 610599 China Southwest University of Science and Technology Mianyang 621010 China Department of Electrical and Computer Engineering University of Alberta Edmonton AB T6G 2V4 Canada Department of Energy Technology Aalborg University Aalborg 9220 Denmark

Energy losses during the conversion and supply of electric power are considered a significant issue and cannot be estimated. Improvement in the efficiency of energy conversion systems is highly restricted because of their internal nonlinearity and complexity. Thus, inspired by the successful utilization of robotic chemists, we demonstrate a pioneering concept of artificial intelligence (AI)-aided automatic online real-time optimization of a power electronics converter using a dual active bridge (DAB) converter as an example. An optimal modulation strategy was obtained through repeated automatic exploration experiments on a practical DAB converter platform. Specifically, the DAB experimental platform operated autonomously around the clock for approximately 71 h. It performed 120,000 consecutive experiments (12,000 episodes) within a six-variable experimental space driven by a deep deterministic policy gradient (DDPG) algorithm. The proposed AI-aided automatic online real-time optimization method achieved significantly improved efficiency of power conversion and supply. Consequently, zero carbon emissions may be obtained in the future. © 2023

关键词： Artificial intelligence automatic online real-time exploration DC-DC converter Dual active bridge converter Energy conversion system Power electronics

来源：评论

学校读者我要写书评

暂无评论

Deep Learning-Based Dimension Measurement System for Industrial Part Inspection

Deep Learning-Based Dimension Measurement System for Industr...

引用

2024 China Automation Congress, CAC 2024

作者： Li, Xiao-Di Fan, Xing-Rong Zhang, Xian-Ming Zhang, Han-Zhi Huang, Chun-Jie Xu, Cheng Chongqing Engineering Laboratory for Detection Control and Integrated System School of Artificial Intelligence Chongqing Technology and Business University Chongqing China Engineering Research Centre for Waste Oil Recovery Technology and Equipment Ministry of Education Chongqing Key Laboratory Chongqing Technology and Business University Chongqing China School of Computer Science and Information Engineering Chongqing Technology and Business University Chongqing China Chongqing Intelligent Supply Chain Engineering & Technology Research Center Chongqing Jiaotong University Chongqing China Chongqing Aerospace Bashan Motorcycle Manufacturing Co. Ltd. Chongqing China

ISBN: (纸本)9798350368604

This paper develops a deep learning-based system for the dimension measurement of industrial parts. The system employs a general self-supervised image segmentation network to perform dichotomous image segmentation (DIS) on the parts, effectively extracting binary images of the target components from complex backgrounds without requiring preprocessing steps such as filtering or denoising. Following segmentation, contour detection, and dimension measurement are executed. The system utilizes a CMOS area scan camera and was tested on a computer equipped with a GTX1050 graphics card. The tests, conducted on two typical thrust ball bearing components (i.e., the ball retainer and the shaft ring) ü yielded ten measurements per part. Comparison with a vernier caliper as the reference standard reveals that under normal operating conditions, the maximum standard deviation is 0.007 mm and the maximum absolute error is 0.017 mm. © 2024 IEEE.

关键词： Ball bearings

来源：评论

学校读者我要写书评

暂无评论

EEG-GNet: GCN-Based EEG Signal Fatigue Detection Network for Edge Devices

EEG-GNet: GCN-Based EEG Signal Fatigue Detection Network for...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Zhang, Fuquan Wu, Yu Xu, Xinxin Wang, Chuansheng Ma, Minjia Grau, Antoni Huang, Jiayan School of Computer and Big Data Minjiang University Fuzhou China Fujian Provincial Key Laboratory of Information Processing and Intelligent Control Minjiang University Fuzhou China College of New Engineering Industry Putian University Putian China Department of Automatic Control Polytechnic University of Catalonia Barcelona Spain School of Robotics Xi'an Jiaotong-Liverpool University Suzhou China College of New Engineering Industry Putian University Fujian Ruituo Information Technology Limited Company Putian China

ISBN: (纸本)9798350386226

Fatigue driving is a common potential traffic safety hazard. To detect and analysis the driver Fatigue degree in real-time, more attention have been increasing paid to application research of EEG fatigue detection device. However, existing methods suffer from complex model structure or large-scale model parameters. In the other words, they requires high computational consumption and large memory footprint, which creates a contradiction between model and portable EEG machine with limited hardware resources. In this work, we take both time and frequency domains of EEG signal into account and design an lightweight and efficient driver fatigue detection network based on GCN (EEG-GNet), which is available edge-end device application. Experiments on 4th Raspberry Pi on a public baseline dataset show that our EEG-GNet outperforms the state-of-the-art methods and takes good tradeoff between detection accuracy and resource consumption. Our EEG-GNet provides a new potential opportunity for future applications in practical portable/wearable (P/W) devices. © 2024 IEEE.

关键词： Electrocardiography

来源：评论

学校读者我要写书评

暂无评论

A Novel Variable Impedance PM Machines with Separated Fault Tolerant Ring

A Novel Variable Impedance PM Machines with Separated Fault ...

引用

IEEE Energy Conversion Congress and Exposition, ECCE

作者： Runyu Wang Xinggang Fan Dawei Li Ronghai Qu Yongtao Peng Xiaopeng Zhao State Key Laboratory of Advanced Electromagnetic Engineering and Technology School of Electrical and Electronic Engineering Huazhong University of Science and Technology Wuhan China National Key Laboratory of Science and Technology on Aircraft Control AVIC Xi’an Flight Automatic Control Research Institute Xi’an China

This paper aims to analyze and design a novel variable impedance permanent magnet (PM) machine with separated fault tolerant ring for high reliability applications. The proposed topology automatically changes from low impedance to ultra-high impedance under turn-to-turn short-circuit without complex fault diagnosis and control strategies. First, the topology of the variable impedance PM machine is introduced. Then, the magnetic field distribution of the separate fault tolerant ring under health and fault is analyzed to qualitatively reveal the fault tolerance principle. Finally, a turn-to-turn short-circuit model considering the core saturation effect is developed for investigating the mechanism of fault tolerance performance improvement. It is found that the short-circuit current and torque ripple after the fault of the variable impedance machines are significantly lower than those of the conventional ones.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Frequency Domain Auto-tuning of Structured LPV controllers for High-Precision Motion control

Frequency Domain Auto-tuning of Structured LPV Controllers f...

引用

IEEE Conference on Decision and control

作者： Yorick Broens Hans Butler Roland Tóth Department of Electrical Engineering Eindhoven University of Technology Eindhoven The Netherlands ASML Veldhoven The Netherlands Systems and Control Laboratory HUNREN Institute for Computer Science and Control Hungary

ISBN: (数字)9798350316339

ISBN: (纸本)9798350316346

Motion systems are a vital part of many industrial processes. However, meeting the increasingly stringent demands of these systems, especially concerning precision and throughput, requires novel control design methods that can go beyond the capabilities of traditional solutions. Traditional control methods often struggle with the complexity and position-dependent effects inherent in modern motion systems, leading to compromises in performance and a laborious task of controller design. This paper addresses these challenges by introducing a novel structured feedback control auto-tuning approach for multiple-input multiple-output (MIMO) motion systems. By leveraging frequency response function (FRF) estimates and the linear-parameter-varying (LPV) control framework, the proposed approach automates the controller design, while providing local stability and performance guarantees. Key innovations include norm-based magnitude optimization of the sensitivity functions, an automated stability check through a novel extended factorized Nyquist criterion, a modular structured MIMO LPV controller parameterization, and a controller discretization approach which preserves the continuous-time (CT) controller parameterization. The proposed approach is validated through experiments using a state-of-the-art moving-magnet planar actuator prototype.

关键词： Technological innovation Sensitivity Frequency-domain analysis Stability criteria Prototypes Throughput Frequency response Motion control Optimization MIMO communication

来源：评论

学校读者我要写书评

暂无评论

Multi-Attention Gate Based U-net for Retinal Vessel Segmentation 12

Multi-Attention Gate Based U-net for Retinal Vessel Segmenta...

引用

12th International Conference on Information technology in Medicine and Education, ITME 2022

作者： Zhu, Yu Li, Rui Teng, Shenghua Cao, Xinrong Shandong University of Science and Technology College of Electronic and Information Engineering Qingdao266590 China College of Computer and Control Engineering Fujian Provincial Key Laboratory of Information Processing and Intelligent Control Minjiang University Fuzhou350121 China

ISBN: (纸本)9798350310153

Retinal blood vessel segmentation images can be used to detect and evaluate various cardiovascular and ophthalmic diseases. However, due to the intricate vessel structures and blurred boundaries of vessels, it is a huge challenge to efficiently and accurately segment blood vessels. To deal with the above problems, this paper improves on the U-net by firstly using multi-scale feature convolution with kernels of varying size for feature extraction. Second, a non-local attention mechanism is applied to obtain richer global semantic information. Then multi-attention gate is used in the skip connection part by inputting feature maps of various scales and dimensions and selectively learning the interrelated regions, which improves the segmentation ability of the network model for the tiny structure of blood vessels. Quantitative and qualitative experimental results on two public datasets, DRIVE and CHASE-DB1, demonstrate the effectiveness of the proposed method. © 2022 IEEE.

关键词： Blood vessels

来源：评论

学校读者我要写书评

暂无评论

A Network Combining CNN and Transformer for Blind Image Super-Resolution 12

A Network Combining CNN and Transformer for Blind Image Supe...

引用

12th International Conference on Information technology in Medicine and Education, ITME 2022

作者： Zhang, Shuhao Li, Zuoyong Teng, Shenghua Zeng, Kun Shandong University of Science and Technology College of Electronic and Information Engineering Qingdao266590 China Minjiang University Fujian Provincial Key Laboratory of Information Processing and Intelligent Control College of Computer and Control Engineering Fuzhou350121 China

ISBN: (纸本)9798350310153

Blind super-resolution (SR) requires not only estimating blur kernel, but also super-resolving low-resolution image based on estimated blur kernel. Most blind SR methods use convolutional neural networks (CNNs) for kernel estimation, which cannot exploit long-range dependency within image domain, thus failing to predict blur kernel accurately. To address this issue, we propose a network combining CNN and transformer named NCCT for kernel estimation. By modeling local and non-local image priors simultaneously, NCCT outperforms other blind SR methods in terms of kernel estimation accuracy. Moreover, we design a network module named RRFDB for constructing lightweight blind SR network, which runs faster and achieves comparative SR performance with fewer parameters compared with other state-of-the-art blind SR methods. © 2022 IEEE.

关键词： Optical resolving power

来源：评论

学校读者我要写书评

暂无评论

On the Approximation of the Quotient of Two Gaussian Densities for Multiple-Model Smoothing 26

On the Approximation of the Quotient of Two Gaussian Densiti...

引用

26th International Conference on Information Fusion, FUSION 2023

作者： Liu, Yi Li, Xi Yang, Le Mihaylova, Lyudmila Xue, Yanbo University of Canterbury Department of Electrical and Computer Engineering Christchurch New Zealand National University of Defense Technology State Key Lab of Complex Electromagnetic Environment Effects on Electronics and Information Systems Changsha China University of Sheffield Department of Automatic Control and System Engineering Sheffield United Kingdom Career Science Lab Beijing China

ISBN: (纸本)9798890344854

The quotient of two multivariate Gaussian densities can be written as an unnormalized Gaussian density, which has been applied in some recently developed multiple-model fixed-interval smoothing algorithms. However, this expression is invalid if instead of being positive definite, the covariance of the unnormalized Gaussian density is indefinite (i.e., it has both positive and negative eigenvalues) or undefined (i.e., computing it requires inverting a singular matrix). This paper considers approximating the quotient of two Gaussian densities in this case using two different approaches to mitigate the caused numerical problems. The first approach directly replaces the indefinite covariance of the unnormalized Gaussian density with a positive definite matrix nearest to it. The second approach computes the approximation through solving, using the natural gradient, an optimization problem with a Kullback-Leibler divergence-based cost function. This paper illustrates the application of the theoretical results by incorporating them into an existing smoothing method for jump Markov systems and utilizing the obtained smoothers to track a maneuvering target. © 2023 International Society of Information Fusion.

关键词： Cost functions

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：