Machine Learning for Cyber Physical System: Advances and Challenges

Foreword
Preface
Contents
1 SMOTE Integrated Adaptive Boosting Framework for Network Intrusion Detection
	1.1 Introduction
	1.2 Literature Study
	1.3 Proposed Method
	1.4 Experimental Setup
		1.4.1 Experimental Data
		1.4.2 Data Preprocessing
		1.4.3 Simulation Environment and Parameter Setting
		1.4.4 Performance Measures
	1.5 Result Analysis
	1.6 Conclusion
	References
2 An In-Depth Analysis of Cyber-Physical Systems: Deep Machine Intelligence Based Security Mitigations
	2.1 Introduction
	2.2 In-Depth Insights into Cyber-Physical Systems
		2.2.1 Key Characteristics of CPS
		2.2.2 Critical Challenges in CPS Landscape
	2.3 WSNs in the Context of CPS
		2.3.1 MAC and Diverse Protocol Adaptations
	2.4 CPS Based Security and Risk Mitigation
		2.4.1 Attack Types
		2.4.2 Swift CPS Forecasting with Machine Intelligence
		2.4.3 Swift CPS Forecasting with DL
	2.5 Experiment and Results
		2.5.1 Benchmark Dataset
		2.5.2 MI Based Classification
	2.6 Conclusions and Future Scope
	References
3 Unsupervised Approaches in Anomaly Detection
	3.1 Introduction
	3.2 Methodology
		3.2.1 Types of Algorithms
		3.2.2 Evaluation Metrics
	3.3 ANN and CNN Models Integrated with SMOTE
		3.3.1 Oversampling Data: SMOTE
		3.3.2 Examples of Using SMOTE with ANN and CNN
	3.4 Unsupervised Learning in Anomaly Detection in Practice
		3.4.1 Method
	3.5 Development Frameworks
	References
4 Profiling and Classification of IoT Devices for Smart Home Environments
	4.1 Introduction
		4.1.1 Motivations
		4.1.2 Contribution
	4.2 Literature Survey
	4.3 Research Gap and Objectives
	4.4 Methodology and Evaluation
		4.4.1 Acquisition Phase
		4.4.2 Sensor Configuration
		4.4.3 Analysis
		4.4.4 Classification
		4.4.5 Action (Prevention and Recovery)
	4.5 System Model
		4.5.1 Attack Model
		4.5.2 Security Analysis
		4.5.3 Machine Learning Models
		4.5.4 Machine Learning Classifiers
		4.5.5 Analysis of ML Models
		4.5.6 Dataset and Feature Selection
		4.5.7 Evaluation
	4.6 Result and Analysis
	4.7 Conclusions, Challenges and Future Work
	References
5 Application of Machine Learning to Improve Safety in the Wind Industry
	5.1 Introduction
	5.2 Literature Review
		5.2.1 Context
		5.2.2 Overview of the Offshore Wind Industry and Its Safety Challenges
		5.2.3 Review of Traditional Safety Management Practices in the Offshore Wind Industry
		5.2.4 Introduction to Machine Learning and Deep Learning Technologies
		5.2.5 Previous Studies on the Application of Machine Learning to Improve Safety in Other Industries
		5.2.6 Application of ML to the Offshore Wind Industry
		5.2.7 Challenges of Using ML in the Offshore Wind Industry
		5.2.8 Traditional Safety Metrics in the Wind Industry
		5.2.9 Limitations and Challenges Associated with Traditional Safety Metrics
		5.2.10 Potential Benefits of Using Machine Learning and Deep Learning in the Wind Industry
		5.2.11 Summary of the Gaps in the Current Literature and the Research Problem
	5.3 Data Processing
		5.3.1 Data Description
		5.3.2 Columns Description
		5.3.3 Undersampling Technique and SMOTE Technique
	5.4 Models
		5.4.1 Machine Learning Models
		5.4.2 Deep Learning Models
	5.5 Analysis and Results
		5.5.1 Machine Learning Models Results with the Original Dataset
		5.5.2 Machine Learning Models Result for Undersampled Dataset
		5.5.3 Machine Learning Models Results for Oversampling (SMOTE) Dataset
		5.5.4 Neural Network Model Results
		5.5.5 Deep Neural Network Results
		5.5.6 Analaysis of Results
	5.6 Conclusion
	References
6 Malware Attack Detection in Vehicle Cyber Physical System for Planning and Control Using Deep Learning
	6.1 Introduction
		6.1.1 Motivation
		6.1.2 Research Contribution
	6.2 Related Work
	6.3 Methodologies
		6.3.1 RF
		6.3.2 AdaBoost
		6.3.3 GBoost
		6.3.4 Bagging
		6.3.5 XGBoost
		6.3.6 CNN
		6.3.7 Proposed Methodology
	6.4 Experimental Setup and Dataset Overview
		6.4.1 Overview of Dataset
		6.4.2 Data Preparation
		6.4.3 Simulation Environment
		6.4.4 Performance Measures
	6.5 Result Analysis
	6.6 Critical Discussion
	6.7 Conclusion and Future Work
	References
7 Unraveling What is at Stake in the Intelligence of Autonomous Cars
	7.1 Introduction
		7.1.1 Autonomous Systems Scenario: Some Mathematical Modeling Techniques for the Dynamics of Cyber-Physical Systems
	7.2 Disentangling the Cognitive Structure on Which Autonomous Systems Are Based: Perelman and Olbrechts-Tyteca\'s Methodology of Argumentation with an Appeal to Reality
		7.2.1 Human Cognitive Linguistic Process
		7.2.2 AI Cognitive Linguistic Process
		7.2.3 Recapping the Approach Discussed in This Section on Cognitive Structure
	7.3 Advanced Driver Assistance Systems (ADAS) Replacing Humans
		7.3.1 Merging Computational and Physical Resources in ADAS
		7.3.2 Structure Behind the ADAS Design
		7.3.3 Logical and Executive Functions in Autonomous Driving Systems: What is at Stake in the Impenetrable Black Box AI
		7.3.4 Recapping the Fundamentals of Advanced Driver Assistance Systems (ADAS) Cognition
	7.4 Rethinking the Interpretive Activity of Cyber-Physical Systems Under a Unifying Context
		7.4.1 Unraveling the Black Box of Logical and Executive Functions of Autonomous Driving Systems
		7.4.2 Recapping About Dealing with the Complexity and Mystique of Black Box AI.
	7.5 ADAS: Learning Coming Out of Integration Between Algorithmic Core and Context
		7.5.1 The ADAS Learning Process: Principles that Organize the ‘Way of Doing’
		7.5.2 Recapping About Learning Accomplished by ADAS
	7.6 Conclusion
	References
8 Intelligent Under Sampling Based Ensemble Techniques for Cyber-Physical Systems in Smart Cities
	8.1 Introduction
	8.2 Cyber Physical System
	8.3 Feature Selection and Hyperparameter Tuning Challenges
		8.3.1 Feature Selection
		8.3.2 Hyperparameter Tuning
	8.4 Proposed Methodology
		8.4.1 Under-Sampling Ensemble Techniques
	8.5 Related Works
	8.6 Experiment Setup and Datasets Descriptions
		8.6.1 System Environment
		8.6.2 Dataset Description
	8.7 Results and Discussion
	8.8 Conclusion
	References
9 Application of Deep Learning in Medical Cyber-Physical Systems
	9.1 Introduction
	9.2 Related Study
	9.3 Proposed Approach
		9.3.1 Mathematical Background
		9.3.2 Optimization Using Adam Optimizer
	9.4 Description of Dataset and Environmental Setup
		9.4.1 Environmental Setup
		9.4.2 Dataset Description
		9.4.3 Data Preprocessing
	9.5 Analysis of Empirical Findings
		9.5.1 Metrics Used in Validation of Model
		9.5.2 Comparative Assessment of Findings
	9.6 Conclusion
	References
10 Risk Assessment and Security of Industrial Internet of Things Network Using Advance Machine Learning
	10.1 Introduction
	10.2 Literature Survey
	10.3 Methodology
		10.3.1 Gradient Boosting Decision Tree
		10.3.2 Gravitational Search Algorithm
		10.3.3 Proposed Method
	10.4 Result Analysis
		10.4.1 Dataset Information
		10.4.2 Result Analysis
	10.5 Conclusion
	References
11 Machine Learning Based Intelligent Diagnosis of Brain Tumor: Advances and Challenges
	11.1 Introduction
	11.2 System Under Study
	11.3 Materials and Methodology
		11.3.1 Dataset
		11.3.2 Proposed ML Based Brain Tumor Classifier
		11.3.3 Preprocessing
		11.3.4 Features Extraction
		11.3.5 ML Based Classifiers
		11.3.6 Radial Basis Functions
		11.3.7 Activation Function
		11.3.8 K-Fold Cross-Validation
	11.4 Analysis of Result
	11.5 Discussion
	11.6 Conclusion
	References
12 Cyber-Physical Security in Smart Grids: A Holistic View with Machine Learning Integration
	12.1 Introduction
	12.2 Background and Fundamentals Component of Smart-Grid
		12.2.1 Advanced Metering Infrastructure
		12.2.2 Operational Technology Component
		12.2.3 Information Technology Component
	12.3 Cyber Security and Cyber Physical System
		12.3.1 Cyber Threat and Cybersecurity
		12.3.2 Cyber Physical System: Smart Grid
	12.4 A Brief Introduction to Machine Learning (ML) and Deep Learning (DL)
		12.4.1 Shallow Learning
		12.4.2 Deep Learning
	12.5 Cybersecurity in Smart Grid
		12.5.1 Cyber Threats in Smart Grid: Smart Grid Devices Vulnerable to Cyber Attacks
		12.5.2 Cyber Threats in Smart Grid: Proactive Measures
	12.6 Application of ML and DL Algorithms in Smart Grid Cybersecurity
	12.7 Challenges and Future Directions
	12.8 Conclusion
	References
13 Intelligent Biometric Authentication-Based Intrusion Detection in Medical Cyber Physical System Using Deep Learning
	13.1 Introduction
		13.1.1 Research Motivation
		13.1.2 Research Contribution
		13.1.3 Organization of Paper
	13.2 Related Works
	13.3 Basic Preliminaries
		13.3.1 Decision Tree
		13.3.2 Random Forest
		13.3.3 Adaptive Boosting
		13.3.4 GBoost
		13.3.5 XGBoost
		13.3.6 CatBoost
	13.4 Proposed Method
		13.4.1 Convolutional Layer
		13.4.2 Pooling Layer
		13.4.3 Fully Connected Layer
		13.4.4 Activation Function
	13.5 Dataset Description and Simulation Setup
		13.5.1 Dataset Description
		13.5.2 Dataset Preprocessing
		13.5.3 Experimental Setup
		13.5.4 Evaluation Measures
	13.6 Result Analysis
	13.7 Conclusions
	References
14 Current Datasets and Their Inherent Challenges for Automatic Vehicle Classification
	14.1 Introduction
		14.1.1 Reason of Interest
	14.2 Categorization of AVC-Based Datasets
		14.2.1 Aerial Image-Based Vehicle Datasets
		14.2.2 Frontal Image-Based AVC Datasets
		14.2.3 Video-Based AVC Datasets
	14.3 Research Gaps and Challenges Limitations Related to AVC Datasets
		14.3.1 Future Scope
	14.4 Conclusion
	References