检索结果-内蒙古大学图书馆

Proceedings of the 37th International Conference on Neural Information Processing Systems

作者： Enrico Giudice Jack Kuipers Giusi Moffa Dep. of Mathematics and Computer Science University of Basel Basel Switzerland Dep. of Biosystems Science and Engineering ETH Zurich Basel Switzerland Dep. of Mathematics and Computer Science University of Basel Basel Switzerland and Division of Psychiatry University College London London UK

Gaussian Process Networks (GPNs) are a class of directed graphical models which employ Gaussian processes as priors for the conditional expectation of each variable given its parents in the network. The model allows the description of continuous joint distributions in a compact but flexible manner with minimal parametric assumptions on the dependencies between variables. Bayesian structure learning of GPNs requires computing the posterior over graphs of the network and is computationally infeasible even in low dimensions. This work implements Monte Carlo and Markov Chain Monte Carlo methods to sample from the posterior distribution of network structures. As such, the approach follows the Bayesian paradigm, comparing models via their marginal likelihood and computing the posterior probability of the GPN features. Simulation studies show that our method outperforms state-of-the-art algorithms in recovering the graphical structure of the network and provides an accurate approximation of its posterior distribution.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Enhancing the Efficiency of Convolutional Neural Networks through Quantization 4

Enhancing the Efficiency of Convolutional Neural Networks th...

引用

4th IEEE International Conference on ICT in Business Industry and Government, ICTBIG 2024

作者： Kansal, Vipashi Shnain, Ammar Hameed Pushkarna, Gaurav Manjunatha Dixit, Krishna Kant Rajkumar, K. Varada Graphic Era Deemed to be University Department of Computer Science & Engg Dehradun India College of Technical Engineering The Islamic University Department of Computers Techniques Engg Najaf Iraq Lovely Professional University Phagwara India New Horizon College of Engg Department of Mechanical Engg Bangalore India Gla University Department of Electrical Engg Mathura India Mlr Institute of Technology Department of CSE-AI&ML Telangana Hyderabad India

ISBN: (纸本)9798331518981

A new kind of image categorisation technology, Convolutional Neural Networks (CNNs) have shown themselves capable of astounding accuracy across a range of uses. Problems arise, however, when dealing with real-time applications in settings where resources are limited due to their computational complexity and resource needs. An extensive investigation on how to make convolutional neural networks (CNNs) better at picture classification is detailed in this work. We explore several optimization techniques, including network pruning, quantization, and the deployment of lightweight architectures such as MobileNet and SqueezeNet. Additionally, we investigate the impact of advanced training strategies like transfer learning and data augmentation on model performance. We show, by means of comprehensive tests, that our suggested approaches considerably reduce the memory footprint and computational cost of CNNs while preserving or even enhancing classification accuracy. Our findings provide valuable insights for deploying efficient CNN models in practical scenarios, paving the way for more accessible and scalable image classification solutions. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Deformable Graph Transformer

引用

IEEE Transactions on Pattern Analysis and Machine Intelligence 2025年第7期PP卷 PP页

作者： Park, Jinyoung Yun, Seongjun Park, Hyeonjin Kang, Jaewoo Jeong, Jisu Kim, Kyung-Min Ha, Jung-Woo Kim, Hyunwoo J. The Department of Computer Science and Engineering Korea University Seoul02841 Korea Republic of Amazon Search AI Seattle98170 United States NAVER NAVER Corp. Bundang13561 Korea Republic of NAVER and NAVER CLOUD NAVER Corp. Bundang13561 Korea Republic of

Transformer-based models have recently shown success in representation learning on graph-structured data beyond natural language processing and computer vision. However, the success is limited to small-scale graphs due to the drawbacks of full dot-product attention on graphs such as the quadratic complexity with respect to the number of nodes and message aggregation from enormous irrelevant nodes. To address these issues, we propose Deformable Graph Transformer (DGT) that performs sparse attention via dynamically selected relevant nodes for efficiently handling large-scale graphs with a linear complexity in the number of nodes. Specifically, our framework first constructs multiple node sequences with various criteria to consider both structural and semantic proximity. Then, combining with our learnable Katz Positional Encodings, the sparse attention is applied to the node sequences for learning node representations with a significantly reduced computational cost. Extensive experiments demonstrate that our DGT achieves superior performance on 7 graph benchmark datasets with 2.5 ∼ 449 times less computational cost compared to transformer-based graph models with full attention. © 1979-2012 IEEE. All rights reserved.

关键词： Undirected graphs

来源：评论

学校读者我要写书评

暂无评论

Development of measurement system for subsurface scattering light of skin and analysis of its age-related changes

Development of measurement system for subsurface scattering ...

引用

Biomedical Imaging and Sensing Conference 2023

作者： Kikuchi, K. Katsuyama, M. Shibata, T. Hardeberg, J.Y. Yuasa, T. aizu, Y. Shiseido Co. Ltd. MIRAI Technology Institute 1-2-11 Takashima Nishiku Yokohama220-0011 Japan Department of Computer Science Faculty of Information Technology and Electrical Engineering Norwegian University of Science and Technology Teknologivegen 22 Gjøvik2815 Norway Division of Production Systems Engineering Muroran Institute of Technology 27-1 Mizumoto Muroran Hokkaido050-8585 Japan

ISBN: (纸本)9781510663435

We have developed a system that separates and measures the optical properties of skin, i.e., the surface reflection, diffuse reflection, and sub-surface scattering components of the skin. This system includes two polarization filters that separate light from the skin into a surface reflection component image and a diffuse reflection component image. Furthermore, by using a projector as a light source and irradiating the skin using a high-frequency binary illumination pattern, the subsurface scattering component image alone can be separated and generated. Using the proposed system, we performed a survey of 154 Japanese women aged from their 20s to their 70s and analyzed age-related changes in the optical properties of their skin. The results revealed the following. First, the luminance value Y of the surface reflection from the cheek and its standard deviation within the analysis area increase with age. Second, the Y value of diffuse reflection from the skin decreases with age. Third, the amount of light in the sub-surface scattering components also decreases with age. The proposed system is expected to have a wide range of applications in the medical and cosmetic fields. © 2023 SPIE.

关键词： Surface scattering

来源：评论

学校读者我要写书评

暂无评论

Adaptive Distributed Target Detection for FDA-MIMO Radar in Clutter Environment without Training data

Adaptive Distributed Target Detection for FDA-MIMO Radar in ...

引用

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Bang Huang Ping Li Jiangwei Jian Wen-Qin Wang Shunsheng Zhang Lei Wu School of Information and Communication Engineering University of Electronic Science and Technology of China Chengdu China Computer Electrical and Mathematical Science and Engineering (CEMSE) Division King Abdullah University of Science and Technology (KAUST) Thuwal Saudi Arabia Institute of Electronic Science and Technology University of Electronic Science and Technology of China Chengdu China

ISBN: (数字)9798350360325

ISBN: (纸本)9798350360332

This paper studies the problem of adaptive distributed target detection for frequency diverse array multiple-input multiple-output (FDA-MIMO) radar, where the target is embedded in Gaussian clutter with unknown covariance matrix. The proposed FDA-MIMO radar detection model establishes distributed target detection as a summation expression, which is different from the traditional detection models in MIMO or phase array radars that discuss only point-like target. Next, according to the rules of generalized likelihood ratio tests (GLRT), Rao and Wald tests, we designed three adaptive detectors without the training data. The numerical results validate the proposed method and all theoretical analyses.

关键词： Adaptation models Radar clutter Radar detection Training data Object detection Detectors Frequency diversity

来源：评论

学校读者我要写书评

暂无评论

TTT: A Temporal Refinement Heuristic for Tenuously Tractable Discrete Time Reachability Problems

arXiv

引用

arXiv 2024年

作者： Sidrane, Chelsea Tumova, Jana The Division of Robotics Perception and Learning Intelligent Systems Department School of Electrical Engineering & Computer Science KTH Royal Institute of Technology Brinellvägen 8 Stockholm114 28 Sweden

Reachable set computation is an important tool for analyzing control systems. Simulating a control system can show that the system is generally functioning as desired, but a formal tool like reachability analysis can provide a guarantee of correctness. For linear systems, reachability analysis is straightforward and fast, but as more complex components are added to the control system such as nonlinear dynamics or a neural network controller, reachability analysis may slow down or become overly conservative. To address these challenges, much literature has focused on spatial refinement, e.g., tuning the discretization of the input sets and intermediate reachable sets. However, this paper addresses a different dimension: temporal refinement. The basic idea of temporal refinement is to automatically choose when along the horizon of the reachability problem to execute slow symbolic queries which incur less approximation error versus fast concrete queries which incur more approximation error. Temporal refinement can be combined with other refinement approaches and offers an additional "tuning knob" with which to trade off tractability and tightness in approximate reachable set computation. Here, we introduce an automatic framework for performing temporal refinement and we demonstrate the effectiveness of this technique on computing approximate reachable sets for nonlinear systems with neural network control policies. We demonstrate the calculation of reachable sets of varying approximation error under varying computational budget and show that our algorithm is able to generate approximate reachable sets with a similar amount of error to the baseline approach in 20-70% less time. © 2024, CC BY.

关键词： Linear systems

来源：评论

学校读者我要写书评

暂无评论

Energy-superconvergent Runge-Kutta Time Discretizations

arXiv

引用

arXiv 2024年

作者： Liu, Jinjie Brio, Moysey Division of Physics Engineering Mathematics and Computer Science Delaware State University DoverDE19901 United States Department of Mathematics The University of Arizona TucsonAZ85721 United States

In this paper, we investigate the energy accuracy of explicit Runge-Kutta (RK) time discretization for antisymmetric autonomous linear systems and present a framework for constructing RK methods with an order of energy accuracy much greater than the number of stages. For an s-stage, pth-order RK method, we show that the energy accuracy can achieve superconvergence with an order up to 2s − p + 1 if p is even. Several energy-superconvergent methods, including five- to seven-stage fourth-order methods with energy accuracy up to the eleventh order, together with their strong stability criteria, are derived. The proposed methods are examined using several applications, including second-order ordinary differential equations for harmonic oscillators, linear integro-differential equations for peridynamics, and one-dimensional Maxwell's equations of *** Codes 65L06, 65M12, 35Q61, 74A70 Copyright © 2024, The Authors. All rights reserved.

关键词： Runge Kutta methods

来源：评论

学校读者我要写书评

暂无评论

An Applied Artificial Intelligence Technique for Early-Stage Alzheimer's Disease Prediction

An Applied Artificial Intelligence Technique for Early-Stage...

引用

International Conference on Emerging Trends in Networks and computer Communications (ETNCC)

作者： Hassan Ali Hussain Imtiaz Attique Ur Rehman Sabeen Javaid Tahir Muhammad Ali Menwa Alshammeri Azka Mir Himanshu Kumar Department of Software Engineering University of Sialkot Sialkot Pakistan Department of Computer Science Gulf University for Science and Technology Kuwait Department of Computer Science College of CIS Al Jouf University Saudia Arabia Division of Research and Development Lovely Professional University Phagwara India KKR and KSR Institute of Technology and Sciences Guntur Andhra Pradesh India

ISBN: (数字)9798350353266

ISBN: (纸本)9798350353273

Alzheimer's disease is a neurodegenerative disorder in which central nervous cells gradually die that poses a significant health challenge globally, especially in older people. As life expectancy increases which suggests a large amount of society will be affected. This increasing number of Alzheimer's cases demands the urgency to develop predictive models for early intervention as the real effect of Alzheimer's shows very late. Although there are no treatments capable of reversing the natural pathological changes, we can delay the development of Alzheimer's. Our goal is to, detect this neurodegenerative disease before it becomes more rooted, we can help the patient identify the disease at an early age so they can adapt to their new condition and perform treatments to help manage symptoms. We used the OASIS dataset which consists of multiple features collection of subjects whose age varies from 18 to 96. This research focuses on employing supervised machine learning, we have used various classifiers such as Decision trees, Naïve Bayes, k-NN, random forest, and logistic regression, and evaluated each classification's effectiveness using several well-known performance indicators to forecast Alzheimer's disease status. We used Rapid Miner to test different models and found that the Naïve Bayes classifier achieves the highest accuracy at 97.5% this model has a superior predictive capability and outperforms alternative classifiers. The insights gained from this research contribute to advancing early diagnosis and intervention for Alzheimer's disease.

关键词： Pathology Logistic regression Computational modeling Predictive models Market research Bayes methods Delays Decision trees Alzheimer's disease Random forests

来源：评论

学校读者我要写书评

暂无评论

Robust practical tracking control of an underactuated hovercraft

Robust practical tracking control of an underactuated hoverc...

引用

作者： Yan, Lixia Ma, Baoli Xie, Wenjing The Seventh Research Division The School of Automation Science and Electrical Engineering Beihang University Beijing China School of Computer and Information Science Southwest University Chongqing China

This paper studies the robust practical tracking control problem of an underactuated hovercraft with unknown external disturbances. Controlling such a system is difficult and challenging due to inherent modeling properties such as underactuated nature and nonholonomic constraint. State transformation technique and cascade method, combined with a transverse function, are used to develop a novel control law that globally practically stabilizes the hovercraft to any smooth reference trajectory, either feasible or non-feasible. The Lyapunov theory is applied to the analysis of overall closed-loop stability. Simulation results illustrate the control design. © 2021 Chinese Automatic Control Society and John Wiley & Sons Australia, Ltd

关键词： Navigation

来源：评论

学校读者我要写书评

暂无评论

Optimizing Rare Disease Patient Matching with Large Language Models

Optimizing Rare Disease Patient Matching with Large Language...

引用

2024 IEEE International Conference on Big Data, BigData 2024

作者： Berger, Armin Bashir, Ali Hamza Berghaus, David Mowmita Afsan, Nazia Grigull, Lorenz Fendrich, Lara Hogl, Henriette Ernst, Gundula Schmidt, Ralf Bascom, David Lagones, Tom Anglim Deuber, Tobias Bell, Thiago Lubbering, Max Sifa, Rafet Fraunhofer Iais Department of Media Engineering Germany University of Bonn Department of Computer Science Germany Federal Ministry of Education and Research West-AI Germany Lamarr Institute Germany University Hospital Bonn Center for Rare Diseases Germany Children's Network for Chronic Illnesses and Disabilities Germany Medical School Hannover Department of Medical Psychology Germany Department of Health Queensland Australia Griffith University - School of Medicine Australia

ISBN: (纸本)9798350362480

We present RepLLaMA, a neural ranking model for optimizing patient matching in rare disease communities. Using data from *** consisting of over two thousand profiles and over ten thousand ratings, our bi-encoder architecture maps profiles to 4096-dimensional vectors, enabling efficient similarity computations. The system processes unstructured symptom descriptions and structured responses, incorporating expert-guided LLM enhancements. Results show Top-10 Recall of 49.36% $(\pm 2.03)$, surpassing baselines while maintaining generalization. The implementation provides a scalable solution for rare disease patient matching, addressing computational complexity challenges. © 2024 IEEE.

关键词： Diseases

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：