检索结果-内蒙古大学图书馆

generative machine learning METHODS FOR MULTIVARIATE ENSEMBLE POSTPROCESSING

ANNALS OF APPLIED STATISTICS 2024年第1期18卷 159-183页

作者： Chen, Jieyu Janke, Tim Steinke, Florian Lerch, Sebastian Karlsruhe Inst Technol Inst Econ Karlsruhe Germany Tech Univ Darmstadt Energy Informat Networks & Syst Darmstadt Germany

Ensemble weather forecasts based on multiple runs of numerical weather prediction models typically show systematic errors and require postprocessing to obtain reliable forecasts. Accurately modeling multivariate dependencies is crucial in many practical applications, and various approaches to multivariate postprocessing have been proposed where ensemble predictions are first postprocessed separately in each margin and multivariate dependencies are then restored via copulas. These two-step methods share common key limitations, in particular, the difficulty to include additional predictors in modeling the dependencies. We propose a novel multivariate postprocessing method based on generative machine learning to address these challenges. In this new class of nonparametric data-driven distributional regression models, samples from the multivariate forecast distribution are directly obtained as output of a generative neural network. The generative model is trained by optimizing a proper scoring rule, which measures the discrepancy between the generated and observed data, conditional on exogenous input variables. Our method does not require parametric assumptions on univariate distributions or multivariate dependencies and allows for incorporating arbitrary predictors. In two case studies on multivariate temperature and wind speed forecasting at weather stations over Germany, our generative model shows significant improvements over state-of-the-art methods and particularly improves the representation of spatial dependencies.

关键词： generative machine learning ensemble postprocessing multivariate postprocessing probabilistic forecasting weather forecasting

来源：评论

学校读者我要写书评

暂无评论

Data augmentation and generative machine learning on the cloud platform

引用

International Journal of Information Technology (Singapore) 2024年第8期16卷 4833-4843页

作者： Vyas, Piyush Ragothaman, Kaushik Muthusamy Chauhan, Akhilesh Rimal, Bhaskar Texas A & amp M University-Central Texas College of Business Administration Killeen TX United States University of Wisconsin–Parkside 900 Wood Rd Kenosha 53144 WI United States Dakota State University College of Business & amp Information Systems Madison SD United States University of Idaho 875 Perimeter Drive MS 1010 Moscow 83844-1010 ID United States

This paper aims to explore the image data augmentation application on the cloud platform utilizing state-of-the-art generative machine learning techniques. This paper further highlights these techniques’ significance in addressing the challenge of data generation and emphasizes the need for further research in this area. This research adopts an in-depth exploration approach to examine the burgeoning domain of generative machine learning techniques. It discusses the evolution of these techniques and their integration with cloud services powered by Graphical Processing Unit (GPU)-enabled computational engines. Practical experimentation involving Modified National Institute of Standards and Technology (MNIST) data is conducted to showcase the capabilities of generative models, with a focus on the core generative Adversarial Network (GAN). The findings reveal the potential of generative machine learning techniques in generating new data images, as demonstrated through practical experimentation with MNIST data. It also highlights the ongoing evolution of these techniques and their challenges, particularly in terms of computational requirements and integration with cloud computing services. This research originally contributes to the existing literature by providing insights into recent advancements and challenges in GANs and their synergies with cloud computing. It presents results from experimentation and emphasizes the importance of cost-effective development environments for implementing generative machine learning techniques. © Bharati Vidyapeeth's Institute of Computer Applications and Management 2024.

关键词： Cloud computing generative adversarial network generative machine learning Google cloud platform Image augmentation

来源：评论

学校读者我要写书评

暂无评论

generative machine learning for de novo drug discovery: A systematic review

引用

COMPUTERS IN BIOLOGY AND MEDICINE 2022年 145卷 105403-105403页

作者： Martinelli, Dominic D. Independent researcher United States of America

Recent research on artificial intelligence indicates that machine learning algorithms can auto-generate novel drug-like molecules. generative models have revolutionized de novo drug discovery, rendering the explorative process more efficient. Several model frameworks and input formats have been proposed to enhance the performance of intelligent algorithms in generative molecular design. In this systematic literature review of experimental articles and reviews over the last five years, machine learning models, challenges associated with computational molecule design along with proposed solutions, and molecular encoding methods are discussed. A query-based search of the PubMed, ScienceDirect, Springer, Wiley Online Library, arXiv, MDPI, bioRxiv, and IEEE Xplore databases yielded 87 studies. Twelve additional studies were identified via citation searching. Of the articles in which machine learning was implemented, six prominent algorithms were identified: long short-term memory recurrent neural networks (LSTM-RNNs), variational autoencoders (VAEs), generative adversarial networks (GANs), adversarial autoencoders (AAEs), evolutionary algorithms, and gated recurrent unit (GRURNNs). Furthermore, eight central challenges were designated: homogeneity of generated molecular libraries, deficient synthesizability, limited assay data, model interpretability, incapacity for multi-property optimization, incomparability, restricted molecule size, and uncertainty in model evaluation. Molecules were encoded either as strings, which were occasionally augmented using randomization, as 2D graphs, or as 3D graphs. Statistical analysis and visualization are performed to illustrate how approaches to machine learning in de novo drug design have evolved over the past five years. Finally, future opportunities and reservations are discussed.

关键词： Artificial intelligence Cheminformatics generative machine learning de novo drug discovery

来源：评论

学校读者我要写书评

暂无评论

Design Ideation with AI - Sketching, Thinking and Talking with generative machine learning Models 23

Design Ideation with AI - Sketching, Thinking and Talking wi...

引用

ACM Designing Interactive Systems Conference (DIS) on Rebuilding and Resilience

作者： Tholander, Jakob Jonsson, Martin Stockholm Univ Stockholm Sweden Sodertorn Univ Huddinge Sweden

ISBN: (纸本)9781450398930

generative machine learning models provide opportunities to support design work in various parts of the design process. This study investigates how generative machine learning and large language models may play a part in creative design processes of ideation, early prototyping and sketching. A workshop was conducted in which design practitioners and design researchers developed design concepts for a provided design case, with the help of GPT-3. The fndings point to three main themes, including i) the practical usefulness and limitations of the system in design ideation processes, ii) how the form of user interaction shapes users' expectations of the system's capabilities and potentials, and iii), how the broader discourse around AI both limits and enables how co-creative processes involving human and AI unfolds. The discussion outlines design implications and alternative framings of this kind of co-creative design practices based on post-human perspectives on design and technology use.

关键词： GPT-3 ChatGPT Large Language Models LLM generative machine learning co-creation post-human design ideation computer supported ideation

来源：评论

学校读者我要写书评

暂无评论

ProcessGAN: Supporting the creation of business process improvement ideas through generative machine learning

引用

DECISION SUPPORT SYSTEMS 2023年 165卷

作者： van Dun, Christopher Moder, Linda Kratsch, Wolfgang Roeglinger, Maximilian Univ Bayreuth FIM Res Ctr Wittelsbacherring 10 D-95444 Bayreuth Germany Fraunhofer FIT Branch Business & Informat Syst Engn Wittelsbacherring 10 D-95444 Bayreuth Germany Univ Appl Sci Augsburg FIM Res Ctr Alter Postweg 101 D-86159 Augsburg Germany Fraunhofer FIT Branch Business & Informat Syst Engn Alter Postweg 101 D-86159 Augsburg Germany

Business processes are a key driver of organizational success, which is why business process improvement (BPI) is a central activity of business process management. Despite an abundance of approaches, BPI as a creative task is time-consuming and labour-intensive. Most importantly, its level of computational support is low. The few computational BPI approaches hardly leverage the opportunities brought about by computational creativity, neglect process data, and rely on rather rigid improvement patterns. Given the increasing amount of process data in the form of event logs and the uptake of generative machine learning for automating creative tasks in various domains, there is huge potential for BPI. Hence, following the design science research paradigm, we specified, implemented, and evaluated ProcessGAN, a novel computational BPI approach based on generative adversarial networks that supports the creation of BPI ideas. Our evaluation shows that ProcessGAN improves the creativity of process designers, particularly the originality of BPI ideas, and shapes up useful in real-world settings. Moreover, ProcessGAN is the first approach to combine BPI and computational creativity.

关键词： Business process improvement Business process redesign generative adversarial networks generative machine learning Process mining

来源：评论

学校读者我要写书评

暂无评论

generative Physics Informed machine learning Method for DC-Link Capacitance Estimation

引用

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 2025年第5期72卷 5461-5471页

作者： Qie, Tianhao Zhang, Xinan Xiang, Chaoqun Zhao, Shuai Jiang, Chaoqiang Iu, Herbert H. C. Fernando, Tyrone Univ Western Australia Dept Elect Elect & Comp Engn Perth WA 6009 Australia Cent South Univ Sch Traff & Transportat Engn Changsha 410083 Peoples R China Aalborg Univ Dept Energy Technol DK-9220 Aalborg Denmark City Univ Hong Kong Dept Elect Engn Hong Kong 999077 Peoples R China

This article proposes an innovative generative physics-informed machine learning (GPIML) method for the estimation of dc-link capacitance during the precharging process of the vehicular power systems, which contributes to greatly enhancing the reliability of electrified transportation. Different from the other machine learning-based estimation approaches, the proposed method produces highly accurate results using small input experimental dataset. To enable sufficient neural network training, diffusion algorithm is first adopted in the proposed method to augment the training data based on small input dataset. Then, the augmented data is fed to a physics-informed long short-term memory (PILSTM) algorithm to estimate the dc-link capacitance. Superior accuracy and strong robustness to measurement noises are achieved. The effectiveness of the proposed method is validated through experimental studies.

关键词： Capacitance Capacitors Estimation machine learning Physics Training Noise Mathematical models Resistance Data models Capacitance estimation dc-link capacitor electrified transportation generative machine learning physic informed machine learning

来源：评论

学校读者我要写书评

暂无评论

Active machine learning for Pre-procedural Prediction of Time-Varying Boundary Condition After Fontan Procedure Using generative Adversarial Networks

引用

ANNALS OF BIOMEDICAL ENGINEERING 2025年第1期53卷 217-229页

作者： Song, Wenyuan Frakes, David Dasi, Lakshmi Prasad Georgia Inst Technol Sch Elect & Comp Engn Atlanta GA USA Apple Inc Cupertino CA USA Georgia Inst Technol Wallace H Coulter Dept Biomed Engn Atlanta GA 30322 USA Emory Univ Atlanta GA 30332 USA

The Fontan procedure is the definitive palliation for pediatric patients born with single ventricles. Surgical planning for the Fontan procedure has emerged as a promising vehicle toward optimizing outcomes, where pre-operative measurements are used prospectively as post-operative boundary conditions for simulation. Nevertheless, actual post-operative measurements can be very different from pre-operative states, which raises questions for the accuracy of surgical planning. The goal of this study is to apply machine leaning techniques to describing pre-operative and post-operative vena caval flow conditions in Fontan patients in order to develop predictions of post-operative boundary conditions to be used in surgical planning. Based on a virtual cohort synthesized by lumped-parameter models, we proposed a novel diversity-aware generative adversarial active learning framework to successfully train predictive deep neural networks on very limited amount of cases that are generally faced by cardiovascular studies. Results of 14 groups of experiments uniquely combining different data query strategies, metrics, and data augmentation options with generative adversarial networks demonstrated that the highest overall prediction accuracy and coefficient of determination were exhibited by the proposed method. This framework serves as a first step toward deep learning for cardiovascular flow prediction/regression with reduced labeling requirements and augmented learning space.

关键词： Congenital heart disease Fontan procedure generative machine learning Active learning Uncertainty measurement Flow prediction Predictive modeling

来源：评论

学校读者我要写书评

暂无评论

Enhancing Brain Age Prediction: A generative AI Approach for EEG machine learning Models

Enhancing Brain Age Prediction: A Generative AI Approach for...

引用

IEEE International Instrumentation and Measurement Technology Conference (I2MTC)

作者： Cook, Zara Sinha, Grant Wang, Jack Zhao, Chengzong Belacel, Nabil Doesburg, Sam Medvedev, George Ribary, Urs Vakorin, Vasily Xi, Pengcheng Simon Fraser Univ Waterloo ON Canada Univ Waterloo Waterloo ON Canada Univ Waterloo Comp Sci Waterloo ON Canada CNR Waterloo ON Canada Univ Waterloo Comp Engn Waterloo ON Canada Natl Res Council Canada Digital Technol Ottawa ON Canada Simon Fraser Univ Biomed Physiol & Kinesiol Burnaby BC Canada Fraser Hlth Author Div Neurol Vancouver BC Canada

ISBN: (纸本)9798350380903;9798350380910

Predicting the age of brain from EEG data holds significant promise for advancing healthcare diagnostics. By comparing predicted brain age with the biological age of patients, clinicians can gain valuable insights into the overall health status of individuals. To assist clinical decision making, the latest advance in AI research shows promise in the analysis of routine clinical data;however, challenges including limited availability of high-quality training data are limiting the full capacity of AI in neurology and clinical workflow. Therefore, this work proposes a novel approach using generative machine learning models to augment the training dataset for building models to predict brain age from EEG recordings. Our findings reveal that integrating synthetic data significantly boosts the performance of models. The study holds significant implications for neurological engineering, in particular for EEG-based age prediction tasks.

关键词： EEG diagnostics generative machine learning data augmentation brain age prediction

来源：评论

学校读者我要写书评

暂无评论

Efficient property-oriented design of composite layups via controllable latent features using generative VAE

引用

COMPOSITES SCIENCE AND TECHNOLOGY 2025年 259卷

作者： Sun, Huaixiang Wang, Xiaodong Li, Jiaxue Li, Zengshan Guan, Zhidong Beihang Univ Sch Aeronaut Sci & Engn Beijing 100191 Peoples R China

Fiber-reinforced composites provide substantial tailoring potential, while the extensive parameters and complex coupling mechanisms pose formidable challenges to layup designs. This paper presents an efficient inverse design framework for composite layups utilizing a variational autoencoder (VAE), which is applicable to nonconventional laminates. By leveraging the VAE's exceptional feature extraction and generative capabilities, the decoder rapidly produces layups with desired properties through controllable feature vectors. Based on the stacking characteristics of layups, multi-scale one-dimensional convolutions precisely extract sequence features relevant to mechanical properties and specific manufacturing constraints. A customized loss function is formulated to constrain the latent features, while addressing the non-uniqueness problem for layups with certain mechanical properties. The developed property-oriented VAE can generate 100,000 layups in seconds, achieving an average success rate of 66.9 % under comprehensive in-plane and bending stiffness design, and remains effective for 100-ply thick laminate. For comparison, the VAE model outperforms the genetic algorithm and the logic-based method in reinforced panel designs, reducing the retrieval error by 46.4 % and 38.1 %, respectively. The proposed approach demonstrates flexible and efficient design advantages using generative machine learning models, and is easily extendable to other inverse design scenarios.

关键词： Polymer-matrix composites (PMCs) Non-conventional laminates Inverse design generative machine learning Latent features

来源：评论

学校读者我要写书评

暂无评论

High-Precision Constitutive Modeling of CMC Interphase Under Thermo-Chemo-Mechanical Conditions Based on Molecular Simulation and machine learning

引用

APPLIED COMPOSITE MATERIALS 2025年第3期32卷 971-993页

作者： Chen, Yixin Chen, Shaohua Li, Shiyao You, Chao Wu, Tao Wang, Fang Xu, Nuo Gao, Xiguang Song, Yingdong Nanjing Univ Aeronaut & Astronaut Coll Energy & Power Engn Nanjing 210016 Peoples R China Minist Educ Key Lab Inlet & Exhaust Syst Technol Nanjing 210016 Peoples R China Nanjing Forestry Univ Coll Chem Engn Nanjing 210037 Peoples R China Hohai Univ Inst Solid Mech Coll Mech & Engn Sci Nanjing 211100 Peoples R China

Ceramic matrix composite (CMC) is emerging as a leading candidate for next-generation aeronautical materials. While ceramics are brittle, CMCs demonstrate improved toughness thanks to the matrix-fiber interphase, which deflects crack propagation. To date, accurately predicting the mechanical behavior of the CMC interphase under complex thermo-chemo-mechanical conditions remains a major challenge. In this context, we introduce an AI-based generative framework that directly generates highly accurate strain-stress relations for the CMC interphase based on measurements of temperature, oxidation state, and strain rate. The model combines an unsupervised autoencoder, which learns the key features of the strain-stress relation, with a multilayer feed-forward neural network that maps loading conditions to these features. Pre-trained by extensive molecular dynamics simulations and calibrated with minimal experimental data, the model is thoroughly validated through push-in tests of single-fiber composites and tensile tests of unidirectional fiber-bundle composites, demonstrating satisfactory accuracy. The primary application of this AI-based method is to evaluate the mechanical performance of the CMC interphase directly from easily measurable loading conditions, bypassing the need for microstructure. This approach offers an efficient solution for load design and health monitoring of ceramic matrix composite structures.

关键词： Ceramic matrix composite Interface/interphase generative machine learning Constitutive modeling Molecular dynamics

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：