Machine Learning in Industry

Preface
Contents
About the Editors
Fundamentals of Machine Learning
	1 Introduction
	2 Artificial Intelligence
	3 Data Analytics
		3.1 Types of Data Analytics
		3.2 Data Mining
	4 Big Data
	5 Supervised Learning
	6 Unsupervised Learning
	7 Reinforcement Learning
	8 Decision Tree
	9 Least Squares
	10 Linear Regression
	11 Neural Networks
	12 Cluster Analysis
	13 Deep Learning
	14 Summary
	References
Neural Network Model Identification Studies to Predict Residual Stress of a Steel Plate Based on a Non-destructive Barkhausen Noise Measurement
	1 Introduction
	2 Model Identification
		2.1 Feature Extraction
		2.2 Feature Selection
		2.3 Neural Network Structure Selection
		2.4 Final Model Training
		2.5 Model Performance Assessment
	3 Considered Data Set
	4 Results and Discussion
		4.1 The Selected Features
		4.2 Final Model Selection and Model Performance Assessment
	5 Concluding Remarks and Recommendations
	References
Data-Driven Optimization of Blast Furnace Iron Making Process Using Evolutionary Deep Learning
	1 Background
		1.1 Challenges in Multi-objective Optimization by Evolutionary Techniques
	2 Blast Furnace
		2.1 Blast Furnace Modeling and Optimization
	3 Evolutionary Data-Driven Models
		3.1 Predator-Prey Genetic Algorithm (PPGA)
		3.2 Evolutionary Neural Network (EvoNN)
		3.3 Bi-Objective Genetic Programming (BioGP)
		3.4 Evolutionary Deep Neural Network (EvoDN2)
		3.5 Reference Vector Evolutionary Algorithm (RVEA)
	4 Handling a Real-World Blast Furnace Iron Making Problem Using Evolutionary Approaches
		4.1 Objective and Scope
		4.2 Data Preparation
		4.3 Time Series Modeling
		4.4 Many Objectives Optimization Formulation
		4.5 Construction of Metamodels
		4.6 Training and the Correlation Coefficients
		4.7 Single Variable Response (SVR)
		4.8 Optimization Work
		4.9 Result Analysis
	5 Conclusion and Future Prospect
	References
A Brief Appraisal of Machine Learning in Industrial Sensing Probes
	1 Introduction
	2 Functional Flowchart of ML
		2.1 Data Pretreatment
		2.2 Pattern Recognition and Dimension Reduction
		2.3 Machine Modeling
	3 Survey
	4 Final Remarks and Recommendations
	References
Mining the Genesis of Sliver Defects Through Rough and Fuzzy Set Theories
	1 Introduction
	2 Dataset
	3 Methodology
		3.1 p-value
		3.2 Rough Set Model
		3.3 Fuzzy Inference System (FIS)
	4 Results and Discussion
		4.1 p-value Results
		4.2 Rough Set Results
		4.3 Development of FIS Model
		4.4 FIS Prediction
	5 Experimental Trials for Validation
	6 Conclusion
	References
Machine Learning Studies in Materials Science
	1 Introduction: The 4th Paradigm of Science in the 4th Industrial Revolution
	2 Methods
	3 Materials–Processes–Knowledge Formalization
	4 Materials
	5 Processes
	6 Formalization of Knowledge: Development of Knowledge Bases and Semantic Integration
	7 Conclusions
	References
Accurate, Real-Time Replication of Governing Equations of Physical Systems with Transpose CNNs — for Industry 4.0 and Digital Twins
	1 Introduction
	2 Related Work and Significance in Industry
		2.1 Related Developments
		2.2 Transpose Convolutions
		2.3 Real-Time Mapping from Sensors to Field Interior Values
	3 Application Domains for Demonstration of Concepts
		3.1 Compressible Potential Flow Over a Flat Plate at Incidence
		3.2 Strong Shocks and Axial Drag Resolution over ONERA-M6 Wing
	4 Principles of CNN Architectures for Relevant Application Domains
		4.1 Architectural Principles
		4.2 Presentation of Results
	5 Conclusions
	References
Deep Learning in Vision-Based Automated Inspection: Current State and Future Prospects
	1 Introduction to Computer Vision and Machine Vision
		1.1 Vision-Based Automated Inspection
	2 Introduction to Deep Learning
		2.1 Traditional Vision Versus Deep Learning for Vision-Based Inspection
	3 Deep Learning Methods in Machine Vision
		3.1 Challenges in Adopting Deep Learning for Machine Vision
		3.2 Enablers of DL in Machine Vision
		3.3 Deep Learning-Based Machine Vision Project Pipeline
		3.4 Deep Learning Hardware
		3.5 Deep Learning Software
	4 Industries Using Deep Learning-Based Machine Vision
	5 Future Prospects
	6 5G Technology
	References
Performance Improvement in Hot Rolling Process with Novel Neural Architectural Search
	1 Introduction
	2 Formulation
		2.1 Evolutionary Neural Architecture Search for Optimal ANN Design
		2.2 Hot Rolling Process Description and Data Generation
	3 Results and Discussions
		3.1 Data Pre-processing
		3.2 Neural Architecture Search
		3.3 Discussions and Future Scope
	4 Conclusions
	References