检索结果-内蒙古大学图书馆

A TRUST-REGION METHOD FOR SOLVING TRUNCATED COMPLEX SINGULAR VALUE DECOMPOSITION

Journal of Computational Mathematics 2024年第4期42卷 999-1031页

作者： Jiaofen Li Lingchang Kong Xuefeng Duan Xuelin Zhou Qilun Luo School of Mathematics and Computing Science Guangxi Colleges and Universities Key Laboratory of Data Analysis and ComputationGuangxi Key Laboratory of Automatic Detecting Technology and InstrumentsGuilin University of Electronic TechnologyCenter for Applied Mathematics of Guangxi(GUET)Guilin 541004China School of Mathematical Sciences South China Normal UniversityGuangzhou 510000China School of Mathematics and Computing Science Guilin University of Electronic TechnologyGuilin 541004China School of Mathematics and Computing Science Guangxi Colleges and Universities Key Laboratory of Data Analysis and ComputationGuilin University of Electronic TechnologyCenter for Applied Mathematics of Guangxi(GUET)Guilin 541004China School of Mathematics and Statistics Yunan UniversityKunming 650000China

The truncated singular value decomposition has been widely used in many areas of science including engineering,and statistics,*** this paper,the original truncated complex singular value decomposition problem is formulated as a Riemannian optimiza-tion problem on a product of two complex Stiefel manifolds,a practical algorithm based on the generic Riemannian trust-region method of Absil et *** presented to solve the underlying problem,which enjoys the global convergence and local superlinear conver-gence *** experiments are provided to illustrate the efficiency of the proposed *** with some classical Riemannian gradient-type methods,the existing Riemannian version of limited-memory BFGS algorithms in the MATLAB toolbox Manopt and the Riemannian manifold optimization library ROPTLIB,and some latest infeasible methods for solving manifold optimization problems,are also provided to show the merits of the proposed approach.

关键词： Truncated singular value decomposition Riemannian optimization Trust-region method

来源：评论

学校读者我要写书评

暂无评论

The implicit bias of adam on separable data 24

The implicit bias of adam on separable data

引用

Proceedings of the 38th International Conference on Neural Information Processing Systems

作者： Chenyang Zhang Difan Zou Yuan Cao Department of Statistics and Actuarial Science School of Computing and Data Science The University of Hong Kong Department of Computer Science School of Computing and Data Science & Institute of Data Science The University of Hong Kong Department of Statistics and Actuarial Science School of Computing and Data Science & Department of Mathematics The University of Hong Kong

ISBN: (纸本)9798331314385

Adam has become one of the most favored optimizers in deep learning problems. Despite its success in practice, numerous mysteries persist regarding its theoretical understanding. In this paper, we study the implicit bias of Adam in linear logistic regression. Specifically, we show that when the training data are linearly separable, the iterates of Adam converge towards a linear classifier that achieves the maximum ℓ∞ -margin in direction. Notably, for a general class of diminishing learning rates, this convergence occurs within polynomial time. Our result shed light on the difference between Adam and (stochastic) gradient descent from a theoretical perspective.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Alignment Work for Urban Accessibility: A Study of How Wheelchair Users Travel in Urban Spaces

引用

Proceedings of the ACM on Human-Computer Interaction 2022年第CSCW2期6卷 1–22页

作者： Wu, Yiying Ding, Xianghua Dai, Xuelan Zhang, Peng Lu, Tun Gu, Ning School of Computer Science Shanghai Key Laboratory of Data Science Fudan University Shanghai China School of Computing Science University of Glasgow Glasgow United Kingdom

Mobility is the key for people with disabilities to have full participation in life. To support their mobility, previous work primarily focused on accessibility as an attribute of the external environment to be evaluated, labeled, visualized, and improved. Relatively less work has looked at how people with disabilities go out and move around in practice and work to achieve mobility. This paper presents a qualitative study of 14 wheelchair users' travel practices in everyday life in China, highlighting urban accessibility as alignment work-the use of an awareness of the conditions along the journey to continuously align and re-Align various resources to ensure accessibility and mobility across contexts and moments. By highlighting alignment work, we present another side of the story of urban accessibility, and broaden design considerations to support mobility for wheelchair users. © 2022 ACM.

关键词： Alignment

来源：评论

学校读者我要写书评

暂无评论

MSNet: Momentum Spillover Network for Indian Stock Market Movement Prediction 20

MSNet: Momentum Spillover Network for Indian Stock Market Mo...

引用

20th IEEE India Council International Conference, INDICON 2023

作者： Patel, Manali Jariwala, Krupa Chattopadhyay, Chiranjoy Svnit Department of Cse Surat India Flame University School of Computing and Data Science Pune India

ISBN: (纸本)9798350305593

Investors and financial analysts require accurate predictions of stock prices to make informed decisions. We present Momentum Spillover Network (MSNet), a deep learning model that predicts stock prices on the National Stock Exchange of India by considering interdependencies existing between stocks in the same sector. Traditional approaches have focused solely on capturing hidden temporal dependencies, ignoring the impact of interconnectedness between stocks. To evaluate MSNet, we experimented with data from companies with the largest market capitalization from various sectors that covers over 70% of the NIFTY-50 index. MSNet effectively identifies the most impactful peer companies and predicts the future price movement of a target company with higher accuracy than existing deep learning approaches, highlighting the importance of incorporating relational dependency between stocks along with temporal dependency. © 2023 IEEE.

关键词： Deep Learning Momentum Spillover Effect Relational Modeling Stock Market Forecasting Temporal Modeling

来源：评论

学校读者我要写书评

暂无评论

Adaptive strategy automation with large language models in paradoxical games

引用

Physical Review Research 2025年第2期7卷 L022012-L022012页

作者： Kang Hao Cheong Jie Zhao Tao Wen Division of Mathematical Sciences School of Physical and Mathematical Sciences Nanyang Technological University Singapore S637371 Singapore College of Computing and Data Science (CCDS) Nanyang Technological University Singapore S639798 Singapore

Finding an ideal sequence in games is crucial for maximizing gains in various scenarios. This process requires extensive investigation, which is both time consuming and demands significant domain expertise. The emergence of large language models (LLMs) represents a significant turning point in addressing this issue, attributable to their potent analytical capabilities. In this context, LLMs can serve to substantially alleviate the human labor needed to manage these complexities. Through comprehensive simulations of coin-tossing games, we have demonstrated that the adaptive switching strategies formulated by LLMs surpass predefined sequences in profitability when applied to certain paradoxical games. Furthermore, our experimental findings indicate that the proposed method not only automates the identification of effective strategies but also provides adaptability to various forms of these paradoxical games.

关键词： Computer simulation languages

来源：评论

学校读者我要写书评

暂无评论

Seg-diffusion: Text-to-Image Diffusion Model for Open-Vocabulary Semantic Segmentation

Seg-diffusion: Text-to-Image Diffusion Model for Open-Vocabu...

引用

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

作者： Zhang, Shuo Huang, Jiaming Wu, Yan Hu, Tao Tang, Wenbing Liu, Jing Shanghai Key Laboratory of Trustworthy Computing East China Normal University Shanghai China Huolala Shenzhen China College of Computing and Data Science Nanyang Technological University Singapore

ISBN: (纸本)9798350368741

Open-vocabulary semantic segmentation (OVSS) is a challenging computer vision task that labels each pixel within an image based on text descriptions. Recent advancements in OVSS are largely attributed to the increased model capacity. However, these models often struggle with unfamiliar images or unseen text, as their visual language understanding is limited to training data. Text-to-image (T2I) diffusion models have demonstrated strong image generation with diverse open-vocabulary descriptions. It prompted us to explore whether the comprehensive priors in T2I diffusion models could enhance the zero-shot generalization of OVSS. In this study, we define OVSS as a denoising diffusion task from noisy to object mask and introduce Seg-diffusion, a novel method based on Stable Diffusion that utilizes its extensive visual and linguistic prior knowledge. Specifically, the object mask diffuses from ground-truth to a random distribution in latent space. The model learns to reverse this noisy process to reconstruct object mask to segment objectives using text embeddings with our proposed Content Attention Module (CAM). Extensive experiments on popular OVSS benchmarks show that Seg-diffusion outperforms previous well-established methods and achieves impressive zero-shot generalization to unseen datasets. © 2025 IEEE.

关键词： content attention module fine-tuning protocol open-vocabulary semantic segmentation text-to-image LDM

来源：评论

学校读者我要写书评

暂无评论

Generative Adversarial Learning for Semi-supervised Retinal Layer Segmentation in OCT Images

Generative Adversarial Learning for Semi-supervised Retinal ...

引用

2024 IEEE-EMBS International Conference on Biomedical and Health Informatics, BHI 2024

作者： Lin Ong, Charlene Zhi Bakar Ali, Asad Abu Rajapakse, Jagath C Msd International GmbH Singapore Singapore College of Computing and Data Science Nanyang Technological University Health Informatics Lab Singapore Singapore

ISBN: (纸本)9798350351552

It is often challenging to obtain large number of labeled data for retinal layer segmentation in optical coherence tomography scans due to the need for expert ophthalmologists. On the other hand, huge quantities of unlabeled scans are often collected in medical centers. In this work, we propose a novel generative adversarial learning framework, GOctSeg, for semi-supervised retinal layer segmentation. GOctSeg consists of a U-Net generator, a discriminator, and a segmentor. The generation of synthetic images from synthetic labels is performed with the U - Net generator, which is used as input together with labeled B-scan patches into the Segmentor network for retinal layer boundary regression. We have also evaluated our methodology on two datasets, the data Resource for Multiple Sclerosis and Healthy Controls and the Duke University Diabetic Macular Edema dataset and demonstrated that our method outperformed or is comparable to other state-of-the-art methods with limited labels for boundary regression. Furthermore, we investigated the performance of GOctSeg on images with low signal-to-noise ratio or with blurred boundaries and showed that our methodology remained robust. Through ablation studies, we demonstrated the utility of synthetic labels for generative learning to guide in semi-supervised retinal layer segmentation. We envision that this methodology can be used to significantly reduce the effort required to obtain labels when there is label scarcity in clinical settings. © 2024 IEEE.

关键词： Ophthalmology

来源：评论

学校读者我要写书评

暂无评论

Evolutionary Neural Architecture Search for 3D Point Cloud Analysis

Evolutionary Neural Architecture Search for 3D Point Cloud A...

引用

2024 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2024

作者： Yang, Yisheng Du, Guodong Toa, Chean Khim Tang, Ho-Kin Goh, Sim Kuan School of Computing and Data Science Xiamen University Malaysia Malaysia China School of Electrical Engineering and Artificial Intelligence Xiamen University Malaysia

ISBN: (纸本)9781665410205

Neural architecture search (NAS) automates neural network design by using optimization algorithms to navigate architecture spaces, reducing the burden of manual architecture design. While NAS has achieved success, applying it to emerging domains, such as analyzing unstructured 3D point clouds, remains underexplored due to the data lying in non-Euclidean spaces, unlike images. This paper presents Success-History-based Self-adaptive Differential Evolution with a Joint Point Interaction Dimension Search (SHSADE-PIDS), an evolution-ary NAS framework that encodes discrete deep neural network architectures to continuous spaces and performs searches in the continuous spaces for efficient point cloud neural architectures. Comprehensive experiments on challenging 3D segmentation and classification benchmarks demonstrate SHSADE-PIDS's capabilities. It discovered highly efficient architectures with higher accuracy, significantly advancing prior NAS techniques. For segmentation on SemanticKITTI, SHSADE-PIDS attained 64.51% mean IoU using only 0.55M parameters and 4.5GMACs, reducing overhead by over 22-26X versus other top methods. For ModelNet40 classification, it achieved 93.4% accuracy with just 1.31M parameters, surpassing larger models. SHSADE-PIDS provided valuable insights into bridging evolutionary algorithms with neural architecture optimization, particularly for emerging frontiers like point cloud learning. © 2024 IEEE.

关键词： Cloud computing architecture

来源：评论

学校读者我要写书评

暂无评论

Optimised Design of a Current Mirror in 150 nm GaAs Technology 5

Optimised Design of a Current Mirror in 150 nm GaAs Technolo...

引用

5th International Conference on computing, Networking, Telecommunications and Engineering sciences Applications, CoNTESA 2024

作者： Qian, Lua Ying Kang, Chia Chao Ding, Tan Jian Bhuiyan, Mohammad Arif Sobhan Minhad, Khairun Nisa Miraz, Mahdi H. Electrical and Electronics Engineering Xiamen University Malaysia Selangor Malaysia School of Computing and Data Science Xiamen University Malaysia Selangor Malaysia

ISBN: (纸本)9798331534530

The Current Mirror (CM) is a basic building block commonly used in analogue and mixed-signal integrated circuits. Its significance lies in its ability to replicate and precisely regulate the current, making it crucial in various applications such as amplifiers, filters and data converters. Recently, there has been a growing need for smaller and more energy-efficient Radio Frequency (RF) devices due to the advancements in wireless communication, the Internet of Things (IoT) and portable electronics. This research aims to propose an improved and optimised CM design focusing on compactness and energy-efficient operation. Through a comprehensive methodology involving transistor sizing, biasing techniques, load resistance selection, frequency response stabilisation and noise analysis, the proposed high swing CM design achieves a gain of at least 6.005 dB, a reduced power consumption of 91.17 mW, a wide bandwidth of 22.60 kHz and improved linearity as well as accuracy through precise current matching and minimised mismatch. This optimised CM design will further boost the realisation of compact and lower power RF devices, contributing to the advancement of analogue circuit design techniques and enhancing system performance, accuracy and reliability. © 2024 IEEE.

关键词： Gallium arsenide

来源：评论

学校读者我要写书评

暂无评论

Reactor field reconstruction from sparse and movable sensors using Voronoi tessellation-assisted convolutional neural networks

引用

Nuclear science and Techniques 2024年第5期35卷 173-185页

作者： He-Lin Gong Han Li Dunhui Xiao Sibo Cheng Paris Elite Institute of Technology Shanghai Jiao Tong UniversityShanghai 200240China School of Mathematical Sciences Tongji UniversityShanghai 200092China Department of Computing Data Science InstituteImperial College LondonLondon SW72AZUK

The aging of operational reactors leads to increased mechanical vibrations in the reactor *** vibration of the incore sensors near their nominal locations is a new problem for neutronic field *** field-reconstruction methods fail to handle spatially moving *** this study,we propose a Voronoi tessellation technique in combination with convolutional neural networks to handle this *** from movable in-core sensors were projected onto the same global field structure using Voronoi tessellation,holding the magnitude and location information of the *** convolutional neural networks were used to learn maps from observations to the global *** proposed method reconstructed multi-physics fields(including fast flux,thermal flux,and power rate)using observations from a single field(such as thermal flux).Numerical tests based on the IAEA benchmark demonstrated the potential of the proposed method in practical engineering applications,particularly within an amplitude of 5 cm around the nominal locations,which led to average relative errors below 5% and 10% in the L_(2) and L_(∞)norms,respectively.

关键词： Voronoi tessellation Field reconstruction Nuclear reactors Reactor physics On-line monitoring

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：