检索结果-内蒙古大学图书馆

41st International Conference on Machine Learning, ICML 2024

作者： Jessica, Loh Sher En Arafat, Naheed Anjum Lim, Wei Xian Chan, Wai Lee Kong, Adams Wai Kin Rolls-Royce@NTU Corporate Lab Nanyang Technological University Singapore College of Computing and Data Science Nanyang Technological University Singapore School of Mechanical & Aerospace Engineering Nanyang Technological University Singapore

Computational fluid dynamics (CFD) simulation is an irreplaceable modelling step in many engineering designs, but it is often computationally expensive. Some graph neural network (GNN)based CFD methods have been proposed. However, the current methods inherit the weakness of traditional numerical simulators, as well as ignore the cell characteristics in the mesh used in the finite volume method, a common method in practical CFD applications. Specifically, the input nodes in these GNN methods have very limited information about any object immersed in the simulation domain and its surrounding environment. Also, the cell characteristics of the mesh such as cell volume, face surface area, and face centroid are not included in the message-passing operations in the GNN methods. To address these weaknesses, this work proposes two novel geometric representations: Shortest Vector (SV) and Directional Integrated Distance (DID). Extracted from the mesh, the SV and DID provide global geometry perspective to each input node, thus removing the need to collect this information through message-passing. This work also introduces the use of Finite Volume Features (FVF) in the graph convolutions as node and edge attributes, enabling its message-passing operations to adjust to different nodes. Finally, this work extends the use of residual training to improve flow field prediction for a GNN scenario with immersed object, when low resolution data is available. Experimental results on two datasets with five different state-of-the-art GNN methods for CFD indicate that SV, DID, FVF and residual training can effectively reduce the predictive error of current GNN-based methods by as much as 41%. Our codes and datasets are available at https://***/toggled/FvFGeo. Copyright 2024 by the author(s)

关键词： Mesh generation

来源：评论

学校读者我要写书评

暂无评论

NeuroPCA: Enhancing Alzheimer's disorder Disease Detection through Optimized Feature Reduction and Machine Learning 3

NeuroPCA: Enhancing Alzheimer's disorder Disease Detection t...

引用

3rd International Conference on Electrical, Electronics, Information and Communication Technologies, ICEEICT 2024

作者： Sagili, Sreenivas Reddy Chidambaranathan, Senthilnathan Nallametti, Nagarjuna Bodele, Harana M. Raja, Laxmi Gayathri, P.G. Rochester Institute of Technology Department of Computer Science United States Virtusa Corporation 20 Corporate Pl South Piscataway United States Cvr College of Engineering Department of Csit Telangana Hyderabad India J D College of Engineering & Management Department of Electronics and Telecommunications Nagpur India Coimbatore India Kongunadu College of Engineering and Technology Department of Artificial Intelligence and Data Science India

ISBN: (纸本)9798350369083

Alzheimer's (memory loss disorder) disorder Disease (AD) poses a significant diagnostic challenge due to its complex etiology and subtle onset of clinical symptoms. The Early and accurate detection will be a crucial for effective management and intervention. The integration of machine Learning-Driven (ML) techniques based on neuro imaging and clinical data may offers a promising avenue for enhancing of AD detection. However, the highest dimensionality of such data could be impede the efficacy and accuracy of ML models due to curse of dimensionality and over fitting. This research will introduces the "NeuroPCA System,"a novelist diagnostic framework by employing Principal Components based Analysis (PCA) for features reduction, tailored specifically for Alzheimer's (memory loss disorder) disorder Disease detection. The NeuroPCA System will systematically reduces the dimensionality of input health data (e.g., MRI, PET scans, and genetic information) while by preserving the most informative variance components crucial for distinguish between AD patients and healthy controls. The proposed methodology may encompasses preprocessing of the multi-modal dataset to normalized feature scales, followed by the application of PCA to extracts the principal components that capturing of the maximum variance in the dataset. The reduced feature set will then utilized to train and validate a suite of machine Learning-Driven models like Supportive Vector Machines (SVM) for classify subjects into AD and non-AD groups. The effectiveness of the NeuroPCA System may evaluated depends on classification accuracy, sensitivity, specificity, and computational efficiency, by using a cross- validated dataset of diagnosed AD patients and a control group. The NeuroPCA System also demonstrates a significant improvement in model performance and computational efficiency, by highlighting the potential of PCA for feature reduction in the context of Alzheimer's (memory loss disorder) disorder Disease detect

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Progressive Alignment with VLM-LLM Feature to Augment Defect Classification for the ASE dataset

Progressive Alignment with VLM-LLM Feature to Augment Defect...

引用

2025 IEEE International Conference on Consumer Electronics, ICCE 2025

作者： Hsu, Chih-Chung Lee, Chia-Ming Yu, Po-Tsun Kang, Cheng-Jun Sun, Chun-Hung Wu, Kuang-Ming Institute of Data Science National Cheng Kung University Taiwan Corporate R&D Advanced Semiconductor Engineering Group Taiwan

ISBN: (纸本)9798331521165

Traditional defect classification approaches face two main challenges: insufficient training data with unstable quality, and over-dependence on visual modality. These issues are particularly problematic when image patterns are monotonic across defect classes. To address these concurrent problems, we propose a novel approach leveraging vision-language models (VLM) and large language models (LLM) for their zero-shot capabilities. We introduce the ASE dataset, which includes rich textual descriptions alongside images, enabling multi-modal defect classification. Our method employs prompting techniques with VLM-LLM to activate extra-modality features, a progressive feature alignment (PFA) block to refine image-text features in few-shot scenarios, and a cross-modality attention fusion (CMAF) module for effective multi-modal feature integration. Experimental results demonstrate the superiority of our approach over existing defect classification methods on the ASE dataset, showcasing its effectiveness in handling complex, multi-modal industrial data with limited samples. © 2025 IEEE.

关键词： Image coding

来源：评论

学校读者我要写书评

暂无评论

Progressive Alignment with VLM-LLM Feature to Augment Defect Classification for the ASE dataset

Progressive Alignment with VLM-LLM Feature to Augment Defect...

引用

IEEE International Conference on Consumer Electronics (ICCE)

作者： Chih-Chung Hsu Chia-Ming Lee Po-Tsun Yu Cheng-Jun Kang Chun-Hung Sun Kuang-Ming Wu Institute of Data Science National Cheng Kung University Taiwan Corporate R&D Advanced Semiconductor Engineering Group Taiwan

ISBN: (数字)9798331521165

ISBN: (纸本)9798331521172

Traditional defect classification approaches face two main challenges: insufficient training data with unstable quality, and over-dependence on visual modality. These issues are particularly problematic when image patterns are monotonic across defect classes. To address these concurrent problems, we propose a novel approach leveraging vision-language models (VLM) and large language models (LLM) for their zero-shot capabilities. We introduce the ASE dataset, which includes rich textual descriptions alongside images, enabling multimodal defect classification. Our method employs prompting techniques with VLM-LLM to activate extra-modality features, a progressive feature alignment (PFA) block to refine image-text features in few-shot scenarios, and a cross-modality attention fusion (CMAF) module for effective multimodal feature integration. Experimental results demonstrate the superiority of our approach over existing defect classification methods on the ASE dataset, showcasing its effectiveness in handling complex, multimodal industrial data with limited samples.

关键词： Training Visualization Large language models Zero shot learning Training data Contrastive learning Transformers data models Faces Consumer electronics

来源：评论

学校读者我要写书评

暂无评论

OCR is All you need: Importing Multi-Modality into Image-based Defect Detection System

arXiv

引用

arXiv 2024年

作者： Hsu, Chih-Chung Lee, Chia-Ming Sun, Chun-Hung Wu, Kuang-Ming Institute of Data Science National Cheng Kung University Taiwan Corporate R&D Advanced Semiconductor Engineering Group Taiwan

Automatic optical inspection (AOI) plays a pivotal role in the manufacturing process, predominantly leveraging high-resolution imaging instruments for scanning purposes. It detects anomalies by analyzing image textures or patterns, making it an essential tool in industrial manufacturing and quality control. Despite its importance, the deployment of models for AOI often faces challenges. These include limited sample sizes, which hinder effective feature learning, variations among source domains, and sensitivities to changes in lighting and camera positions during imaging. These factors collectively compromise the accuracy of model predictions. Traditional AOI often fails to capitalize on the rich mechanism-parameter information from machines or inside images, including statistical parameters, which typically benefit AOI classification. To address this, we introduce an external modality-guided data mining framework, primarily rooted in optical character recognition (OCR), to extract statistical features from images as a second modality to enhance performance, termed OANet (Ocr-Aoi-Net). A key aspect of our approach is the alignment of external modality features, extracted using a single modality-aware model, with image features encoded by a convolutional neural network. This synergy enables a more refined fusion of semantic representations from different modalities. We further introduce feature refinement and a gating function in our OANet to optimize the combination of these features, enhancing inference and decision-making capabilities. Experimental outcomes show that our methodology considerably boosts the recall rate of the defect detection model and maintains high robustness even in challenging scenarios. © 2024, CC BY.

关键词： Sensitivity analysis

来源：评论

学校读者我要写书评

暂无评论

Progressive Alignment with VLM-LLM Feature to Augment Defect Classification for the ASE dataset

arXiv

引用

arXiv 2024年

作者： Hsu, Chih-Chung Lee, Chia-Ming Sun, Chun-Hung Wu, Kuang-Ming Institute of Data Science National Cheng Kung University Taiwan Corporate R&D Advanced Semiconductor Engineering Group Taiwan

Traditional defect classification approaches are facing with two barriers. (1) Insufficient training data and unstable data quality. Collecting sufficient defective sample is expensive and time-costing, consequently leading to dataset variance. It introduces the difficulty on recognition and learning. (2) Over-dependence on visual modality. When the image pattern and texture is monotonic for all defect classes in a given dataset, the performance of conventional AOI system cannot be guaranteed. A main question is,"how to solve those two problems when they occur at the same time?" The feasible strategy is to explore another feature within dataset and combine an eminent vision-language model (VLM) and Large-Language model (LLM) with their astonishing zero-shot capability. In this work, we propose the special ASE dataset, including rich data description recorded on image, for defect classification, but the defect feature is uneasy to learn directly. Secondly, We present the prompting for VLM-LLM against defect classification with the proposed ASE dataset to activate extra-modality feature from images to enhance performance. Then, We design the novel progressive feature alignment (PFA) block to refine image-text feature to alleviate the difficulty of alignment under few-shot scenario. Finally, the proposed Cross-modality attention fusion (CMAF) module can effectively fuse different modality feature. Experiment results have demonstrated our method’s effectiveness over several defect classification methods for the ASE dataset. © 2024, CC BY.

关键词： Defects

来源：评论

学校读者我要写书评

暂无评论

OCR is All you need: Importing Multi-Modality into Image-based Defect Classification System

OCR is All you need: Importing Multi-Modality into Image-bas...

引用

IEEE Asia-Pacific Conference on Circuits and Systems

作者： Chih-Chung Hsu Chia-Ming Lee Po-Tsun Yu Chun-Hung Sun Kuang-Ming Wu Institute of Data Science National Cheng Kung University Tainan Taiwan Corporate R&D Advanced Semiconductor Engineering Group Kaoshuing Taiwan

ISBN: (数字)9798350378771

ISBN: (纸本)9798350378788

Automatic optical inspection (AOI) plays a pivotal role in the manufacturing process. It detects anomalies by analyzing image patterns. Despite its importance, the deployment of models for AOI often faces challenges. These include limited sample sizes, variations among dataset, and camera-poses. Besides, AOI often fails to capture a rich mechanism-parameter information from machines or inside images, which typically benefit AOI classification. To address this, we introduce an external modality-guided defect classification framework, rooted in optical character recognition (OCR), to extract features from images as a external modality to enhance performance, termed OANet (Ocr-Aoi-Net). A key aspect of our approach is the feature alignment between differemt modality features, extracted using a single modality-aware model. We further introduce feature refinement and a gating function in our OANet to optimize the combination of these features, enhancing inference and decision-making capabilities. Experimental outcomes show that our methodology considerably imporoved the defect classification model and maintains robustness even in challenging scenarios.

关键词： Measurement Manufacturing processes Perturbation methods Optical character recognition Semantics Feature extraction Optical imaging Robustness data mining Resilience

来源：评论

学校读者我要写书评

暂无评论

Advancing Strategic Planning and Dynamic Control of Complex Projects

arXiv

引用

arXiv 2024年

作者： Teuber, L.G. van Heukelum, H.J. Wolfert, A.R.M. Data Science & Engineering Boskalis Rosmolenweg 20 Zuid-Holland Papendrecht3356 LK Netherlands Corporate R&D Boskalis Rosmolenweg 20 Zuid-Holland Papendrecht3356 LK Netherlands

Strategic project planning and dynamic control is essential for ensuring complex projects to be executed best-fit for common purpose. When best-for-project strategies are no longer conceivable for humans, there is a need for effective and efficient computer-aided decision support. To this end, standard simulation-driven evaluation and a-posteriori decision-making is often used instead of a combined simulation and optimisation approach that a-priori integrates the stakeholders goal orientation and risk management into a best-fit design solution. However, recently developed state-of-the-art planning and control methodologies that do already incorporate this integration still lack a stochastic representation and/or an associative multi-objective approach. This paper presents a new project management methodology, called Odycon (Open design and dynamic control), which dissolves the aforementioned shortcomings. To enable this, a generic mathematical statement is first formulated for project planning and dynamic control uniting technical (logistical) capabilities, human goal-oriented behaviour, and stakeholders interests. Then, both Monte-Carlo simulation (MCS) and the Integrative Maximisation of Aggregated Preferences (IMAP) optimisation method are combined, resulting in a best-fit strategic planning and dynamic control methodology. Odycon's use and added value are demonstrated for two applications: (1) a pure strategic planning of an offshore wind installation project, and (2) a pure dynamic control of a highway infrastructure construction project. Both applications demonstrate Odycon's advances in concurrent and associative design and decision-making, offering best-fit for common purpose synthesis for different complex project phases. © 2024, CC BY-NC-ND.

关键词： Stochastic systems

来源：评论

学校读者我要写书评

暂无评论

Finite volume features, global geometry representations, and residual training for deep learning-based CFD simulation 24

Finite volume features, global geometry representations, and...

引用

Proceedings of the 41st International Conference on Machine Learning

作者： Loh Sher En Jessica Naheed Anjum Arafat Wei Xian Lim Wai Lee Chan Adams Wai Kin Kong Rolls-Royce@NTU Corporate Lab Nanyang Technological University Singapore and College of Computing and Data Science Nanyang Technological University Singapore Rolls-Royce@NTU Corporate Lab Nanyang Technological University Singapore Rolls-Royce@NTU Corporate Lab Nanyang Technological University Singapore and School of Mechanical & Aerospace Engineering Nanyang Technological University Singapore

Computational fluid dynamics (CFD) simulation is an irreplaceable modelling step in many engineering designs, but it is often computationally expensive. Some graph neural network (GNN)- based CFD methods have been proposed. However, the current methods inherit the weakness of traditional numerical simulators, as well as ignore the cell characteristics in the mesh used in the finite volume method, a common method in practical CFD applications. Specifically, the input nodes in these GNN methods have very limited information about any object immersed in the simulation domain and its surrounding environment. Also, the cell characteristics of the mesh such as cell volume, face surface area, and face centroid are not included in the message-passing operations in the GNN methods. To address these weaknesses, this work proposes two novel geometric representations: Shortest Vector (SV) and Directional Integrated Distance (DID). Extracted from the mesh, the SV and DID provide global geometry perspective to each input node, thus removing the need to collect this information through message-passing. This work also introduces the use of Finite Volume Features (FVF) in the graph convolutions as node and edge attributes, enabling its message-passing operations to adjust to different nodes. Finally, this work extends the use of residual training to improve flow field prediction for a GNN scenario with immersed object, when low resolution data is available. Experimental results on two datasets with five different state-of-the-art GNN methods for CFD indicate that SV, DID, FVF and residual training can effectively reduce the predictive error of current GNN-based methods by as much as 41%. Our codes and datasets are available at https://***/toggled/FvFGeo

关键词：

来源：评论

学校读者我要写书评

暂无评论

Machine Learning with Physics Knowledge for Prediction: A Survey

arXiv

引用

arXiv 2024年

作者： Watson, Joe Song, Chen Weeger, Oliver Gruner, Theo Le, An T. Hansel, Kay Hendawy, Ahmed Arenz, Oleg Trojak, Will Cranmer, Miles D’Eramo, Carlo Bülow, Fabian Goyal, Tanmay Peters, Jan Hoffman, Martin W. Department of Computer Science Technical University of Darmstadt Germany Germany Germany Center for Artificial Intelligence and Data Science University of Würzburg Germany Centre for Cognitive Science Technical University of Darmstadt Germany ABB Corporate Research Center Mannheim Germany Department of Mechanical Engineering Technical University of Darmstadt Germany IBM Research UKI United Kingdom Data Intensive Science University of Cambridge United Kingdom

This survey examines the broad suite of methods and models for combining machine learning with physics knowledge for prediction and forecast, with a focus on partial differential equations. These methods have attracted significant interest due to their potential impact on advancing scientific research and industrial practices by improving predictive models with small- or large-scale datasets and expressive predictive models with useful inductive biases. The survey has two parts. The first considers incorporating physics knowledge on an architectural level through objective functions, structured predictive models, and data augmentation. The second considers data as physics knowledge, which motivates looking at multi-task, meta, and contextual learning as an alternative approach to incorporating physics knowledge in a data-driven fashion. Finally, we also provide an industrial perspective on the application of these methods and a survey of the open-source ecosystem for physics-informed machine learning. Copyright © 2024, The Authors. All rights reserved.

关键词： Prediction models

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：