10734nam 22004693 450 991086108870332120240519090252.03-031-53092-6(MiAaPQ)EBC31343964(Au-PeEL)EBL31343964(CKB)32063341400041(EXLCZ)993206334140004120240519d2024 uy 0engurcnu||||||||txtrdacontentcrdamediacrrdacarrierMultimodal and Tensor Data Analytics for Industrial Systems Improvement1st ed.Cham :Springer International Publishing AG,2024.©2024.1 online resource (388 pages)Springer Optimization and Its Applications Series ;v.2113-031-53091-8 Intro -- Preface -- Contents -- Introduction to Multimodal and Tensor Data Analytics -- 1 Overview -- References -- Part I Functional Methods for Multimodal Data -- Functional Methods for Multimodal Data Analysis -- 1 What Is Functional Data and FDA? -- 1.1 Functional Data -- 1.2 Examples of Functional Data Analyses -- 1.2.1 Functional Principal Component Analysis (FPCA) -- 1.2.2 Functional Linear Model (FLM) -- 2 Why Is FDA Useful for Multimodal Data Analysis? -- 3 How Can FDA Be Used in Multimodal Data Analysis? -- 3.1 Problem Setting -- 3.2 Variations Within Each Unit: FPCA -- 3.3 Variations Between Units: Kernel Method -- 3.4 Estimation and Prediction -- 4 Concluding Remarks -- References -- Advanced Data Analytical Techniques for Profile Monitoring -- 1 Introduction -- 2 Background -- 3 High-Dimensional Profile Monitoring -- 3.1 Cluster-Correlated Profile Data -- 3.2 Sparse-Correlated Profile Data -- 4 In-Profile Monitoring -- 5 Multi-stage Profile Monitoring -- 6 Summary -- References -- Statistical Process Monitoring Methods Based on FunctionalData Analysis -- 1 Introduction -- 2 Overview of Statistical Methods for SPM of Functional Data -- 2.1 Functional Clustering -- 2.1.1 Raw Data Clustering Method -- 2.1.2 Filtering Clustering Methods -- 2.1.3 Adaptive Clustering Methods -- 2.1.4 Distance-Based Clustering Methods -- 2.2 Functional Regression Control Chart -- 3 A Case Study in the Ship CO2 Emission Monitoring -- 4 Conclusions -- References -- Part II Tensor Analytics Methods for Multimodal Data -- Tensor and Multimodal Data Analysis -- 1 Introduction -- 1.1 Notations and Organization -- 2 Preliminaries -- 2.1 Basics of Tensor -- 2.2 CP Decomposition -- 2.3 Tucker Decomposition -- 2.4 Tensor Normal Distribution -- 3 Model-Based Tensor Analysis -- 3.1 Tensor Response Regression -- 3.1.1 Model Formulation -- 3.1.2 Existing Literature.3.2 Tensor Predictor Regression -- 3.2.1 Model Formulation -- 3.2.2 Existing Literature -- 3.3 Tensor Classification -- 3.3.1 Tensor Classification Without Additional Covariates -- 3.3.2 Tensor Classification with Additional Covariates -- 3.4 Tensor Clustering -- 3.4.1 Decomposition-Based Tensor Clustering -- 3.4.2 Probabilistic Model-Based Tensor Clustering -- 3.5 Additional Tensor Models -- 4 Model-Free Tensor Analysis -- 4.1 Tensor Principal Component Analysis -- 4.2 Tensor Dimension Folding -- 4.2.1 Dimension Folding -- 4.2.2 Dimension Folding for Regression Mean -- 4.3 Coupled Decomposition -- 4.3.1 Couple Matrix and Tensor Factorization -- 4.3.2 Coupled Tensor Factorization -- 5 Generalized Liquid Association Analysis -- 5.1 Review of the Liquid Association and Related Extensions -- 5.2 Generalized Liquid Association Analysis -- 5.3 Example: Multimodal PET Analysis -- References -- Tensor Data Analytics in Advanced Manufacturing Processes -- 1 Introduction -- 2 Current Progress of Tensor Data Analytics in Advanced Manufacturing -- 2.1 Images/Videos in Advanced Manufacturing -- 2.2 Point Clouds in Advanced Manufacturing -- 3 Robust Tensor Decomposition for Metal Additive Manufacturing -- 3.1 Smooth and Sparse Tensor Competition -- 3.2 Smooth Sparse Robust Tensor Decomposition -- 4 Conclusion and Future Perspectives -- References -- Part III Spatio-temporal Analytics Methods for Multimodal Data -- Spatiotemporal Data Analysis: A Review of Techniques, Applications, and Emerging Challenges -- 1 Introduction -- 1.1 Spatiotemporal Data -- 1.2 Prevalence of Spatiotemporal Data -- 1.3 Properties of Spatiotemporal Data -- 1.3.1 Spatial Reference -- 1.3.2 Temporal Reference -- 1.3.3 Autocorrelation -- 1.3.4 Heterogeneity -- 1.4 Types of Spatiotemporal Data -- 1.4.1 Event Data -- 1.4.2 Trajectory Data -- 1.4.3 Point Reference Data -- 1.4.4 Raster Data.1.5 Instances and Formats of Spatiotemporal Data -- 2 Review of Techniques Utilized in ST Data Analysis -- 2.1 Traditional Machine Learning Techniques -- 2.1.1 Support Vector Machines (SVM) -- 2.1.2 Random Forest -- 2.1.3 K-Nearest Neighbors (KNN) -- 2.1.4 Decision Trees (DT) -- 2.1.5 DBSCAN -- 2.2 Traditional Machine Learning vs Deep Learning -- 2.3 Commonly Used Deep Learning Algorithms -- 2.3.1 Convolutional Neural Networks -- 2.3.2 Recurrent Neural Networks -- 2.3.3 Autoencoders -- 2.3.4 Generative Adversarial Networks -- 2.3.5 Restricted Boltzmann Machines -- 2.3.6 Graph Convolutional Neural Networks -- 3 Spatiotemporal Data Analysis Tasks -- 3.1 Predictive Learning -- 3.1.1 Events -- 3.1.2 Time Series -- 3.1.3 Spatial Maps and ST Raster -- 3.1.4 Trajectory -- 3.2 Estimation and Inference -- 3.2.1 Spatial Maps -- 3.2.2 Trajectory -- 3.3 Classification -- 3.3.1 Time Series -- 3.3.2 Spatial Maps -- 3.3.3 ST Raster -- 3.4 Anomaly Detection -- 3.4.1 Events -- 3.4.2 Spatial Map -- 3.4.3 Time Series -- 4 Major Application Areas -- 4.1 Transportation and Logistics -- 4.2 Meteorology and Environment -- 4.3 Human and Animal Mobility -- 4.4 Criminal Activity -- 4.5 Healthcare -- 5 A Case Study of a Spatiotemporal Track Association Algorithm -- 5.1 Problem Description -- 5.2 Understanding the Spatiotemporal Data -- 5.3 Proposed Deep Learning Architecture -- 5.4 Results and Discussion -- 5.5 Conclusion of the Case Study -- 6 Spatiotemporal Data Analysis: Challenges and Future Directions -- 7 Conclusion -- References -- Offshore Wind Energy Prediction Using Machine Learning with Multi-Resolution Inputs -- 1 Introduction -- 2 Wind Energy Forecasting: A Multi-Modal Perspective -- 2.1 Problem Definition: Preliminaries and Notation -- 2.2 Case Study: Wind Forecasting for the Offshore Wind Energy Areas in the US NY/NJ Bight.3 A Physics-Guided ML Approach with Multi-Resolution Inputs -- 3.1 AIRU-WRF: An ML Approach for Probabilistic OSW Forecasting -- 3.2 A Simple Approach to Integrating Multi-Resolution Inputs -- 4 Results -- 5 Conclusion and Future Directions -- References -- Sparse Decomposition Methods for Spatio-Temporal AnomalyDetection -- 1 Overview of Anomaly Detection by Decomposition Methods -- 2 Smooth Sparse Decomposition -- 2.1 Formulation -- 2.2 Algorithm for SSD -- 2.3 Properties of SSD, Convergence Rate, and Decomposability -- 2.4 Noiseless Case -- 2.5 Noisy Case -- 2.6 SSD and Robust Statistics -- 3 Sparse Decomposition with Application to Tensor and Spatio-temporal Data -- 3.1 Spatio-temporal Smooth Sparse Decomposition -- 3.2 SSD-Tensor Algorithm for Hotspot Detection -- 3.3 Hotspot Detection by Sparse Decomposition with Random Background -- 3.4 Smooth Robust Tensor Decomposition and Completion -- 4 Sparse Decomposition with Adaptive Sampling Algorithm -- 5 Deep Sparse Decomposition Method -- 5.1 Deep Spatio-temporal Sparse Decomposition for Anomaly Detection -- 5.2 RGI: Robust GAN Inversion for Mask-Free Image Inpainting and Unsupervised Pixel-Wise Anomaly Detection -- 6 Statistical Transfer Learning for Anomaly Detection -- 6.1 Cold-Start Process -- 6.2 Transfer Learning -- 6.3 Formulation -- 6.4 Transferability -- 7 Future Direction for SSD -- References -- Part IV Deep Learning Methods for Multimodal Data -- Multimodal Deep Learning -- 1 Neural Network-Based Fusion -- 1.1 Deep Early Fusion -- 1.2 Deep Intermediate Fusion -- 1.3 Late Fusion in Deep Multimodal Learning -- 1.3.1 Deep Late Fusion Objectives -- 1.4 Generalization -- 2 Applications -- 2.1 Precision Medicine -- 2.2 Autonomous Driving -- 3 Conclusion -- References -- Multimodal Deep Learning for Manufacturing Systems: Recent Progress and Future Trends -- 1 Introduction.2 Introduction to Application Scenarios -- 3 Multimodal Deep Learning for Material Fracture Prediction -- 3.1 Problem Formulation -- 3.2 StressNet-Multimodal Data Fusion for Stress Prediction -- 3.2.1 Model Structure -- 3.2.2 Loss Function -- 3.3 Experiment Results -- 4 Multi-view Graph Convolutional Neural Network for 3D Point Cloud Analytics -- 4.1 Problem Formulation -- 4.2 Surface Defect Identification with Multi-view Fusion over 3D Point Cloud -- 4.2.1 Unsupervised Defect Detection -- 4.2.2 Multi-view Graph Convolutional Neural Network for Defect Classification -- 4.3 Experiment Results -- 4.3.1 Defect Detection Results -- 4.3.2 Defect Classification Results and Selection of Distance Measure -- 5 Multimodal Deep Learning for Assembly Process Optimization -- 5.1 Problem Formulation -- 5.2 Proposed Method -- 5.2.1 Multi-layer Deep Gaussian Process (MGP) and Hyperparameter Estimation -- 5.2.2 Constrained Bayesian Optimization -- 5.2.3 Properties of MGP-CBO -- 5.3 Experiment Results -- 6 Conclusion and Future Directions -- Appendix 1: Proof of Proposition 1 wang2023mvgcn -- Appendix 2: 2-D Constrained Branin-Hoo Function Gelbart2014BayesianOW -- Appendix 3: Preliminaries -- Convolutional Layer -- Graph Convolutional Neural Network -- Bayesian Optimization -- References -- Part V Integration of Domain Knowledge and Multimodal Data -- Synergy of Engineering and Statistics: Multimodal Data Fusion for Quality Improvement -- 1 Two Fundamental Data Fusion Approaches for Quality Improvement -- 2 Fusion of Engineering and Statistics in Modeling and Analysis of Multistage Manufacturing Processes -- 2.1 Stream of Variation Modeling and Analysis of MMP -- 2.2 Causation Modeling and Analysis of MMP -- 3 Knowledge Fusion for Structured High-Dimensional Data -- 3.1 Introduction to Unsupervised Fine-Grained Defect Detection.3.2 Statistical Prior Guided Decomposition-Based Methods for Expert Knowledge Fusion.Springer Optimization and Its Applications SeriesGaw Nathan1739283Pardalos Panos M318341Gahrooei Mostafa Reisi1739284MiAaPQMiAaPQMiAaPQBOOK9910861088703321Multimodal and Tensor Data Analytics for Industrial Systems Improvement4163260UNINA