Machine Learning in Chemical Safety and Health: Fundamentals with Applications

Cover
Title Page
Copyright Page
Contents
List of Contributors
Preface
Chapter 1 Introduction
1.1 Background
1.2 Current State
1.2.1 Flammability Characteristics Prediction Using Quantitative Structure–Property Relationship
1.2.2 Consequence Prediction Using Quantitative Property–Consequence Relationship
1.2.3 Machine Learning in Process Safety and Asset Integrity Management
1.2.4 Machine Learning for Process Fault Detection and Diagnosis
1.2.5 Intelligent Method for Chemical Emission Source Identification
1.2.6 Machine Learning and Deep Learning Applications in Medical Image Analysis
1.2.7 Predictive Nanotoxicology: Nanoinformatics Approach to Toxicity Analysis of Nanomaterials
1.2.8 Machine Learning in Environmental Exposure Assessment
1.2.9 Air Quality Prediction Using Machine Learning
1.3 Software and Tools
1.3.1 R
1.3.2 Python
References
Chapter 2 Machine Learning Fundamentals
2.1 What Is Learning?
2.1.1 Machine Learning Applications and Examples
2.1.2 Machine Learning Tasks
2.2 Concepts of Machine Learning
2.3 Machine Learning Paradigms
2.4 Probably Approximately Correct Learning
2.4.1 Deterministic Setting
2.4.2 Stochastic Setting
2.5 Estimation and Approximation
2.6 Empirical Risk Minimization
2.6.1 Empirical Risk Minimizer
2.6.2 VC-dimension Generalization Bound
2.6.3 General Loss Functions
2.7 Regularization
2.7.1 Regularized Loss Minimization
2.7.2 Constrained and Regularized Problem
2.7.3 Trade-off Between Estimation and Approximation Error
2.8 Maximum Likelihood Principle
2.8.1 Maximum Likelihood Estimation
2.8.2 Cross Entropy Minimization
2.9 Optimization
2.9.1 Linear Regression: An Example
2.9.2 Closed-form Solution
2.9.3 Gradient Descent
2.9.4 Stochastic Gradient Descent
References
Chapter 3 Flammability Characteristics Prediction Using QSPR Modeling
3.1 Introduction
3.1.1 Flammability Characteristics
3.1.2 QSPR Application
3.1.2.1 Concept of QSPR
3.1.2.2 Trends and Characteristics of QSPR
3.2 Flowchart for Flammability Characteristics Prediction
3.2.1 Dataset Preparation
3.2.2 Structure Input and Molecular Simulation
3.2.3 Calculation of Molecular Descriptors
3.2.4 Preliminary Screening of Molecular Descriptors
3.2.5 Descriptor Selection and Modeling
3.2.6 Model Validation
3.2.6.1 Model Fitting Ability Evaluation
3.2.6.2 Model Stability Analysis
3.2.6.3 Model Predictivity Evaluation
3.2.7 Model Mechanism Explanation
3.2.8 Summary of QSPR Process
3.3 QSPR Review for Flammability Characteristics
3.3.1 Flammability Limits
3.3.1.1 LFLT and LFL
3.3.1.2 UFLT and UFL
3.3.2 Flash Point
3.3.3 Auto-ignition Temperature
3.3.4 Heat of Combustion
3.3.5 Minimum Ignition Energy
3.3.6 Gas-liquid Critical Temperature
3.3.7 Other Properties
3.4 Limitations
3.5 Conclusions and Future Prospects
References
Chapter 4 Consequence Prediction Using Quantitative Property–Consequence Relationship Models
4.1 Introduction
4.2 Conventional Consequence Prediction Methods
4.2.1 Empirical Method
4.2.2 Computational Fluid Dynamics (CFD) Method
4.2.3 Integral Method
4.3 Machine Learning and Deep Learning-Based Consequence Prediction Models
4.4 Quantitative Property–Consequence Relationship Models
4.4.1 Consequence Database
4.4.2 Property Descriptors
4.4.3 Machine Learning and Deep Learning Algorithms
4.5 Challenges and Future Directions
References
Chapter 5 Machine Learning in Process Safety and Asset Integrity Management
5.1 Opportunities and Threats
5.2 State-of-the-Art Reviews
5.2.1 Artificial Neural Networks (ANNs)
5.2.2 Principal Component Analysis (PCA)
5.2.3 Genetic Algorithm (GA)
5.3 Case Study of Asset Integrity Assessment
5.4 Data-Driven Model of Asset Integrity Assessment
5.4.1 Condition Monitoring Data Collection
5.4.2 Data Processing and Storage
5.4.3 Data Mining for Risk Quantification and Monitoring Control
5.4.4 AIM Application
5.4.5 The Application of the Framework
5.5 Conclusion
References
Chapter 6 Machine Learning for Process Fault Detection and Diagnosis
6.1 Background
6.2 Machine Learning Approaches in Fault Detection and Diagnosis
6.3 Supervised Methods for Fault Detection and Diagnosis
6.3.1 Neural Network
6.3.1.1 Neural Network Theory and Algorithm
6.3.1.2 Neural Network Learning for Fault Classification
6.3.1.3 Algorithm for Fault Classification Using Neural Network
6.3.2 Support Vector Machine
6.3.2.1 Support Vector Machine Theory and Algorithm
6.3.3 Support Vector Machine Model Selection and Algorithm
6.3.4 Support Vector Machine Multiclass Classification
6.4 Unsupervised Learning Models for Fault Detection and Diagnosis
6.4.1 K-Nearest Neighbors
6.4.2 One-Class Support Vector Machine
6.4.3 One-Class Neural Network
6.4.4 Comparison Between Deep Learning with Machine Learning in Fault Detection and Diagnosis
6.5 Intelligent FDD Using Machine Learning
6.5.1 Model Development
6.5.2 Data Collection
6.5.2.1 Model Development Steps
6.5.2.2 Result Comparison
6.6 Concluding Remarks
References
Chapter 7 Intelligent Method for Chemical Emission Source Identification
7.1 Introduction
7.1.1 Development of Detecting Gas Emission
7.1.2 Development of Source Term Identification
7.2 Intelligent Methods for Recognizing Gas Emission
7.2.1 Leakage Recognition of Sequestrated CO2 in the Atmosphere
7.2.1.1 Gas Leakage Recognition for CO2 Geological Sequestration
7.2.1.2 Case Studies for CO2 Recognition
7.2.2 Emission Gas Identification with Artificial Olfactory
7.2.2.1 Features of Responses in AOS
7.2.2.2 Support Vector Machine Models for Gas Identification
7.2.2.3 Deep Learning Models for Gas Identification
7.3 Intelligent Methods for Identifying Emission Sources
7.3.1 Source Estimation with Intelligent Optimization Method
7.3.1.1 Principle of Source Estimation with Optimization Method
7.3.1.2 Case Studies of Source Estimation with Optimization Method
7.3.2 Source Estimation with MRE-PSO Method
7.3.2.1 Principle of PSO-MRE for Source Estimation
7.3.2.2 Case Studies
7.3.3 Source Estimation with PSO-Tikhonov Regulation Method
7.3.3.1 Principle of PSO-Tikhonov Regularization Hybrid Method
7.3.3.2 Case Study
7.3.4 Source Estimation with MCMC-MLA Method
7.3.4.1 Forward Gas Dispersion Model Based on MLA
7.3.4.2 Source Estimation with MCMC-MLA Method
7.3.4.3 Case Study
7.4 Conclusions and Future Work
7.4.1 Conclusions
7.4.2 Limitations and Future Work
References
Chapter 8 Machine Learning and Deep Learning Applications in Medical Image Analysis
8.1 Introduction
8.1.1 Machine Learning in Medical Imaging
8.1.2 Deep Learning in Medical Imaging
8.2 CNN-Based Models for Classification
8.2.1 ResNet50
8.2.2 YOLOv4 (Darknet53)
8.2.3 Grad-CAM
8.3 Case Study
8.3.1 Background
8.3.2 Study Design
8.3.3 Training and Testing Database Preparation
8.3.4 Results
8.3.4.1 Classification Performance of the Modified ResNet50 Model
8.3.4.2 Classification Performance of the YOLOv4 Model
8.3.4.3 Post-Processing Via Grad-CAM Model and HSV
8.3.5 Conclusion
8.4 Limitations and Future Work
References
Chapter 9 Predictive Nanotoxicology: Nanoinformatics Approach to Toxicity Analysis of Nanomaterials
9.1 Predictive Nanotoxicology
9.1.1 Introduction
9.1.2 Nano Quantitative Structure–Activity Relationship (QSAR)
9.1.3 Importance of Data for Nanotoxicology
9.2 Machine Learning Modeling for Predictive Nanotoxicology
9.2.1 Overview
9.2.2 Unsupervised Learning
9.2.2.1 Data Exploration Via Self-Organizing Maps (SOMs)
9.2.2.2 Evaluating Associations among Sublethal Toxicity Responses
9.2.3 Supervised Learning
9.2.3.1 Random Forest Models
9.2.3.2 Support Vector Machines
9.2.3.3 Bayesian Networks
9.2.3.4 Supervised Classification and Regression-Based Models for Nano-(Q)SARs
9.2.4 Predictive Nano-(Q)SARs for the Assessment of Causal Relationships
9.3 Development of Machine Learning Based Models for Nano-(Q)SARs
9.3.1 Overview
9.3.1.1 Data-Driven Models
9.3.1.2 Mechanistic/Theoretical Models
9.3.2 Data Generation, Collection, and Preprocessing
9.3.3 Descriptor Selection
9.3.4 Model Selection and Training
9.3.5 Model Validation
9.3.5.1 Descriptor Importance
9.3.5.2 Applicability Domain
9.3.6 Model Diagnosis and Debugging
9.4 Nanoinformatics Approaches to Predictive Nanotoxicology
9.5 Summary
References
Chapter 10 Machine Learning in Environmental Exposure Assessment
10.1 Introduction
10.2 Environmental Exposure Modeling
10.3 Machine Learning Exposure Models
10.4 Model Evaluation
10.5 Case Study
10.6 Other Topics
10.6.1 Bias and Fairness
10.6.2 Wearable Sensors
10.6.3 Interpretability
10.6.4 Extreme Events
10.7 Conclusion
References
Chapter 11 Air Quality Prediction Using Machine Learning
11.1 Introduction
11.2 Air Quality and Climate Data Acquisition
11.2.1 Earth Satellite Observation Datasets
11.2.1.1 Basics of Earth Satellite Observations
11.2.1.2 Earth Satellite Products
11.2.2 Ground-Based In Situ Observation Datasets
11.2.2.1 Basics of the Ground-Based In Situ Observations
11.2.2.2 Ground-Based In Situ Products
11.3 Applications of Machine Learning in Air Quality Study
11.3.1 Shallow Learning
11.3.2 Deep Learning
11.4 An Application Practice Example
11.4.1 Satellite Data Acquisition and Variable Selections
11.4.2 Machine Learning and Deep Learning Algorithms
References
Chapter 12 Current Challenges and Perspectives
12.1 Current Challenges
12.1.1 Data Development and Cleaning
12.1.2 Hardware Issues
12.1.3 Data Confidentiality
12.1.4 Other Challenges
12.2 Perspectives
12.2.1 Real-Time Monitoring and Forecast of Chemical Hazards
12.2.2 Toolkits for Dummies
12.2.3 Physics-Informed Machine Learning
References
Index
EULA