Artificial Intelligence for Materials Science (Springer Series in Materials Science, 312)

Preface
Contents
Brief Introduction of the Machine Learning Method
	1 Introduction
	2 Notation
	3 Support Vector Machine
	4 Clustering
		4.1 Expectation-Maximization Algorithm
		4.2 EM Related to K-Means
	5 Regression
		5.1 Linear Regression
		5.2 Polynomial Regression
		5.3 Ridge Regression and LASSO Regression
	6 Decision Tree and Ensemble Learning
	7 Neural Network
	8 Data Set
	9 Conclusion
	References
Machine Learning for High-Entropy Alloys
	1 Overview of High-Entropy Alloys
		1.1 Fundamentals of High-Entropy Alloys
			1.1.1 High-Entropy Effect
			1.1.2 Sluggish Diffusion Effect
			1.1.3 Severe Lattice Distortion Effect
			1.1.4 Cocktail Effect
		1.2 Preparation Methods of High-Entropy Alloys
			1.2.1 Arc Melting
			1.2.2 Mechanical Alloying
			1.2.3 Sputter Deposition
		1.3 Excellent Properties of High-Entropy Alloys
			1.3.1 Hardness
			1.3.2 Strength and Ductility Synergy
			1.3.3 High-Temperature Properties
	2 Computational Modelling and Machine Learning Approaches for High-Entropy Alloys
		2.1 Conventional Computational Methods for HEA Design
			2.1.1 DFT Calculation
			2.1.2 MD Simulation
			2.1.3 CALPHAD Modelling
		2.2 Machine Learning Predictions of Phase Formations
			2.2.1 Single-/Multi-Phase Solid Solution
			2.2.2 Solid-Solution, Intermetallic, and Amorphous Phases
			2.2.3 Precipitation
		2.3 Machine Learning Predictions of Mechanical Properties
			2.3.1 Hardness
			2.3.2 Young\'s Modulus
			2.3.3 Elastic Constants
	3 Summary
	References
Two-Way TrumpetNets and TubeNets for Identification of Material Parameters
	1 Introduction
	2 Problem Setting and Solving Strategy
		2.1 Production of Composites
		2.2 Strategy for Solving Inverse Problems
			2.2.1 Measurement Strategy
			2.2.2 Sensitivity Analysis
			2.2.3 Inverse Analysis Using Two-Way NN Models
	3 Two-Way TrumpetNets and TubeNets
		3.1 Setting Up Forward or Inverse Problem
		3.2 Direct-Weight-Inversion (DWI) Approach
			3.2.1 DWI Formulae for TrumpetNets
			3.2.2 DWI Formulae for TubeNets
	4 Identification of Material Parameters for Composites
		4.1 Setup of the Forward Problem
		4.2 Sensitivity Analysis
		4.3 Case 1: Training Inverse TrumpetNet Using Datasets Generated by FEM Model
		4.4 Case 2: Training Inverse TrumpetNet Using Datasets Generated by the Trained Forward TrumpetNet
		4.5 Case 3: Inverse TrumpetNet Using DWI Approach Without Trained
		4.6 Inverse TubeNet Using DWI Approach
			4.6.1 TubeNets with PCA-Treated Training Samples
			4.6.2 Case 4: Inverse TubeNet with 4-4-4-4 Neurons
			4.6.3 Case 5: Inverse TubeNet with 8-8-8-8 Neurons
	5 Conclusions
	References
Machine Learning Interatomic Force Fields for Carbon Allotropic Materials
	1 Introduction
	2 Traditional Force Fields
	3 Machine Learning Force Field
	4 Procedure to Develop ML-Based Force Field
		4.1 Reference Database
		4.2 Structural Fingerprints
		4.3 Sampling and Clustering
		4.4 Machine Learning
			4.4.1 Kernel Ridge Regression
			4.4.2 Linear Model
			4.4.3 Neural Network Model
	5 Applications of ML Force Fields for Carbon Allotropes
		5.1 ML Force Field for Graphene
		5.2 ML Force Field for Diamond
		5.3 ML Force Field for Amorphous Carbon
	6 Future Directions and Perspective
	References
Genetic Algorithms
	1 Artificial Bee Colony Algorithms
		1.1 Main Steps of the Artificial Bee Colony Algorithm
		1.2 Applications of the Artificial Bee Colony in Functional Materials
		1.3 Current Status of Research on Artificial Bee Colony Algorithms
	2 Particle Swarm Optimization Algorithms
		2.1 Applications of the Particle Swarm Optimization in Multifunctional Materials
		2.2 Current State of the Particle Swarm Optimization
	3 Differential Evolution Algorithms
		3.1 Brief Introduction of the Differential Evolution Algorithm
			3.1.1 Mutation
			3.1.2 Crossover
			3.1.3 Selection
			3.1.4 Other Variants of DE
		3.2 Applications of the Differential Evolution Algorithm
	4 Conclusions
	References
Accelerated Discovery of Thermoelectric Materials Using Machine Learning
	1 Introduction
	2 Methodology
	3 Ingredients for Statistical Methods
	4 High-Throughput Screening and Machine Learning Prediction Models for Thermoelectrics
	5 Connection Between Electronic and Thermal Transport
	6 Towards Universal Models: Algorithm Development
	7 Conclusions and Future Perspective
	References
Thermal Nanostructure Design by Materials Informatics
	1 Introduction
	2 MI Algorithm
		2.1 Bayesian Optimization
		2.2 Monte Carlo Tree Search
		2.3 Shotgun Transfer Learning
	3 MI-Enabled Thermal Conduction Modulation
		3.1 Optimal Nanostructure Design
		3.2 Experimental Realization of Optimal Nanostructure
		3.3 Thermoelectric Nanostructures Design
		3.4 Disordered Structural Parameters Exploring in Aperiodic Superlattices
	4 MI-Enabled Thermal Emission Modulation
		4.1 Thermal Nanostructure Designs for Radiative Cooling
		4.2 Thermal Nanostructures Design for Ultranarrow Thermal Emission
		4.3 Thermal Nanostructure Designs for Thermophotovoltaic System
		4.4 Thermal Nanostructure Designs for Thermal Camouflage
	5 Summary and Perspectives
	References
Machine Learning Accelerated Insights of Perovskite Materials
	1 Introduction
	2 Learning with Perovskite Databases
	3 Materials Representations
		3.1 Descriptors for Perovskites in Photovoltaic Applications
		3.2 Feature Engineering
	4 Machine Learning in Perovskite-Based Material Discovery and Study
		4.1 Stability
		4.2 Photovoltaic Property
	5 Conclusion and Prospects
	References
Index