Advances in Data Clustering: Theory and Applications

Preface
Acknowledgment
Contents
1 Classification of Gougerot-Sjögren Syndrome Based on Artificial Intelligence
	1.1 Introduction
	1.2 Segmentation Process in GSID
	1.3 HarmonicSS Database
	1.4 Data Pre-processing
	1.5 Handcrafted Features: Normalization and Selection
	1.6 Deep Neural Network on GSS Detection
		1.6.1 Weight Initialization
		1.6.2 Segmentation
		1.6.3 Multiphase Joint Training Scheme
		1.6.4 Ydnet Architecture
		1.6.5 Loss Function and Hyperparameters
		1.6.6 Simulations
	1.7 Application on an External Database: HarmonicSS
	1.8 Conclusion
	References
2 Deep Learning Classification of Venous Thromboembolism Based on Ultrasound Imaging
	2.1 Introduction
	2.2 Our Proposed Approach
		2.2.1 Concatenation of Image and Clinical Data
		2.2.2 Architecture Details
		2.2.3 Loss Function and Hyperparameters
	2.3 Simulation Results
		2.3.1 Database Preprocessing
		2.3.2 Segmentation
		2.3.3 Detection of PE with Handcrafted Features
		2.3.4 Detection of Pulmonary Embolism with Deep Learning Models
			2.3.4.1 Clinical Data Fusion
			2.3.4.2 Variation of the Model Width and Depth on DB1
			2.3.4.3 Various Kernel Sizes and Learning Rates for PE Detection
			2.3.4.4 Various Activation Functions and Normalizations for PE Detection
			2.3.4.5 Various Optimizers and Test of Fransfer Learning for PE Detection
		2.3.5 Classification of Recurrent (VTE) with Deep Learning Models
	2.4 Conclusion
	References
3 Synchronization-Driven Community Detection: Dynamic Frequency Tuning Approach
	3.1 Introduction
	3.2 Modeling Community Structures in Networked Systems
	3.3 Network Dynamics
	3.4 Dynamic Tuning Approach
		3.4.1 The Main Algorithm
		3.4.2 From Time Series to Similarity Graph
		3.4.3 Optimal Network Partitioning
	3.5 Experimental Setup
		3.5.1 Network Selection
		3.5.2 Dynamics of the Rössler Oscillators
	3.6 Numerical Results
	3.7 Conclusions
	References
4 Automatic Evolutionary Clustering for Human ActivityDiscovery
	4.1 Introduction
	4.2 Human Activity Discovery Using Clustering
		4.2.1 Preprocessing and Feature Extraction
		4.2.2 Particle Swarm Optimization (PSO)
		4.2.3 Automatic Multi-Objective Clustering Based on Game Theory
		4.2.4 Results and Discussion
	4.3 Other Clustering Techniques
	4.4 Other Unsupervised (Non-clustering) HAR Techniques
	4.5 Conclusion
	References
5 Identification of Correlated Factors for Absenteeism of Employees Using Clustering Techniques
	5.1 Introduction
	5.2 Definition of Clustering
	5.3 Clustering Techniques
		5.3.1 Distribution-Based Clustering
		5.3.2 Density-Based Clustering
		5.3.3 Partition-Based Clustering
		5.3.4 Hierarchical-Based Clustering
		5.3.5 Fuzzy-Based Clustering
		5.3.6 Categorization of Model-Based Clustering
		5.3.7 Grid-Based Clustering
	5.4 Related Works
		5.4.1 Data Set Details
	5.5 Methods and Methodology
		5.5.1 K-Means Algorithm
	5.6 Result Analysis and Conclusion
	References
6 Multi-view Data Clustering Through Consensus Graph and Data Representation Learning
	6.1 Introduction
	6.2 Related Work
		6.2.1 Notations
		6.2.2 Related Work
	6.3 Proposed Approach
	6.4 Optimization of the Proposed MCGLSR (Eq.(6.6))
		6.4.1 Computational Complexity
	6.5 Performance Evaluation
		6.5.1 Experimental Setup
		6.5.2 Experimental Results
		6.5.3 Parameter Sensitivity
		6.5.4 Analysis of Results and Method Comparison
		6.5.5 Convergence Study
	6.6 Conclusion
	References
7 Uber\'s Contribution to Faster Deep Learning: A Case Study in Distributed Model Training
	7.1 Introduction to Distributed Model Training
		7.1.1 Definition of Distributed Model Training
		7.1.2 Benefits of Distributing the Training Process
			7.1.2.1 Parallelization for Speed
			7.1.2.2 Scalability for Big Data
			7.1.2.3 Complexity Unleashed
			7.1.2.4 Beyond Speed: Efficiency and Reliability
			7.1.2.5 Convergence: The Key to Efficiency
	7.2 The HOROVOD Library
		7.2.1 Features of HOROVOD
		7.2.2 Functionalities of HOROVOD
	7.3 Case Study: Uber\'s Contribution
		7.3.1 Specific Case Study Details
		7.3.2 Implementation of Distributed Model Training
		7.3.3 Practical Example: Using HOROVOD
			7.3.3.1 Installation
			7.3.3.2 Setting Up HOROVOD
			7.3.3.3 Example Usage
		7.3.4 Scientific and Technical Aspects
		7.3.5 Challenges and Solutions
		7.3.6 Results and Impact
	7.4 Conclusion
	References
8 Auto-weighted Multi-view Clustering with Unified Binary Representation and Deep Initialization
	8.1 Introduction
	8.2 Related Work
	8.3 The Proposed Approach
		8.3.1 Anchor-Based Representation
		8.3.2 Common Discrete Representation
		8.3.3 Sample View Auto-weighting
		8.3.4 Binary Matrix Factorization and Overall Objective Function
		8.3.5 Optimization
		8.3.6 Binary Clustering Initialization
	8.4 Performance Evaluation
		8.4.1 Experimental Setup
			8.4.1.1 Datasets
			8.4.1.2 Evaluation Metrics and Competitors
		8.4.2 Parameter Sensitivity
		8.4.3 Computational Complexity
		8.4.4 Ablation Study
		8.4.5 Clustering Initialization Analysis
		8.4.6 Convergence Analysis and Effect of the Number of Anchors
		8.4.7 Comparison with State-of-the-Art Multi-view  Methods
	8.5 Conclusion
	References
9 Clustering with Adaptive Unsupervised Graph Convolution Network
	9.1 Introduction
	9.2 Related Work
	9.3 Proposed GCN-Based Clustering
		9.3.1 Notations
		9.3.2 Model Architecture
		9.3.3 Similarity Matrix Output
		9.3.4 Unsupervised Learning Loss
			9.3.4.1 Deep Kernel k-Means Loss
			9.3.4.2 Spectral Clustering Loss
			9.3.4.3 Global Loss
		9.3.5 Spectral Clustering Loss with an Adaptive Fused  Graph
			9.3.5.1 Adaptive Fused Graph: Automatic and Adaptive α
		9.3.6 Final Learning and Clustering
	9.4 Experiments
		9.4.1 Datasets
		9.4.2 Baselines
		9.4.3 Evaluation Metrics and Experimental Setup
		9.4.4 Performance Evaluation
		9.4.5 Ablation Study
		9.4.6 Performance Comparison for Fixed Graph Fusion Versus Adaptive and Automatic Graph Fusion
		9.4.7 Impact of λ Hyperparameter
	9.5 Discussion
	9.6 Conclusion
	References
10 Graph-Based Semi-supervised Learning for Multi-view Data Analysis
	10.1 Introduction
	10.2 Related Work
		10.2.1 Notations
		10.2.2 Gaussian Field and Harmonic Functions zhu03GFHF
		10.2.3 Local and Global Consistency Zhou03LGC
		10.2.4 Review on Flexible Manifold Embedding Nie10FME
		10.2.5 Data Smoothness Assumption
	10.3 Proposed Approach
		10.3.1 Equally Weighted Model
			10.3.1.1 Learning Model
			10.3.1.2 Optimization
		10.3.2 View-Weighted Model
			10.3.2.1 Learning Model
			10.3.2.2 Optimization
	10.4 Experimental Results
		10.4.1 Datasets
		10.4.2 Image Descriptors
		10.4.3 Data Visualization
		10.4.4 Small Databases
		10.4.5 Large Databases
	10.5 Conclusion
	References
11 Advancements in Fuzzy Clustering Algorithms for Image Processing: A Comprehensive Review and Future Directions
	11.1 Introduction
	11.2 Fuzzy Clustering Algorithms
	11.3 Applications in Image Segmentation
	11.4 Comparative Analysis and Future Directions
	11.5 Conclusions
	References