Deep Learning for Radar and Communications Automatic Target Recognition.

Deep Learning for Radar and Communications Automatic Target Recognition
Contents
Foreword
Preface
CHAPTER 1 Machine Learning and Radio Frequency: Past, Present, and Future
1.1 Introduction
1.1.1 Radio Frequency Signals
1.1.2 Radio Frequency Applications
1.1.3 Radar Data Collection and Imaging
1.2 ATR Analysis
1.2.1 ATR History
1.2.2 ATR from SAR
1.3 Radar Object Classification: Past Approach
1.3.1 Template-Based ATR
1.3.2 Model-Based ATR
1.4 Radar Object Classification: Current Approach
1.5 Radar Object Classification: Future Approach
1.5.1 Data Science
1.5.2 Artificial Intelligence
1.6 Book Organization
1.7 Summary
References
CHAPTER 2 Mathematical Foundations for Machine Learning
2.1 Linear Algebra
2.1.1 Vector Addition, Multiplication, and Transpose
2.1.2 Matrix Multiplication
2.1.3 Matrix Inversion
2.1.4 Principal Components Analysis
2.1.5 Convolution
2.2 Multivariate Calculus for Optimization
2.2.1 Vector Calculus
2.2.2 Gradient Descent Algorithm
2.3 Backpropagation
2.4 Statistics and Probability Theory
2.4.1 Basic Probability
2.4.2 Probability Density Functions
2.4.3 Maximum Likelihood Estimation
2.4.4 Bayes' Theorem
2.5 Summary
References
CHAPTER 3 Review of Machine Learning Algorithms
3.1 Introduction
3.1.1 ML Process
3.1.2 Machine Learning Methods
3.2 Supervised Learning
3.2.1 Linear Classifier
3.2.2 Nonlinear Classifier
3.3 Unsupervised Learning
3.3.1 K-Means Clustering
3.3.2 K-Medoid Clustering
3.3.3 Random Forest
3.3.4 Gaussian Mixture Models
3.4 Semisupervised Learning
3.4.1 Generative Approaches
3.4.2 Graph-Based Methods
3.5 Summary
References
CHAPTER 4 A Review of Deep Learning Algorithms
4.1 Introduction
4.1.1 Deep Neural Networks
4.1.2 Autoencoder.

4.2 Neural Networks
4.2.1 Feed Forward Neural Networks
4.2.2 Sequential Neural Networks
4.2.3 Stochastic Neural Networks
4.3 Reward-Based Learning
4.3.1 Reinforcement Learning
4.3.2 Active Learning
4.3.3 Transfer Learning
4.4 Generative Adversarial Networks
4.5 Summary
References
CHAPTER 5 Radio Frequency Data for ML Research
5.1 Introduction
5.2 Big Data
5.2.1 Data at Rest versus Data in Motion
5.2.2 Data in Open versus Data of Importance
5.2.3 Data in Collection versus Data from Simulation
5.2.4 Data in Use versus Data as Manipulated
5.3 Synthetic Aperture Radar Data
5.4 Public Release SAR Data for ML Research
5.4.1 MSTAR: Moving and Stationary Target Acquisition and Recognition Data Set
5.4.2 CVDome
5.4.3 SAMPLE
5.5 Communication Signals Data
5.5.1 RF Signal Data Library
5.5.2 Northeastern University Data Set RF Fingerprinting
5.6 Challenge Problems with RF Data
5.7 Summary
References
CHAPTER 6 Deep Learning for Single-Target Classification in SAR Imagery
6.1 Introduction
6.1.1 Machine Learning SAR Image Classification
6.1.2 Deep Learning SAR Image Classification
6.2 SAR Data Preprocessing for Classification
6.3 SAR Data Sets
6.3.1 MSTAR SAR Data Set
6.3.2 CVDome SAR Data Set
6.4  Deep CNN Learning
6.4.1 DNN Model Design
6.4.2 Experimentation: Training and Verification
6.4.3  Evaluation: Testing and Validation
6.4.4 Confusion Matrix Analysis
6.5 Summary
References
CHAPTER 7 Deep Learning for Multiple Target Classification in SAR Imagery
7.1 Introduction
7.2 Challenges with Multiple-Target Classification
7.2.1 Constant False Alarm Rate Detector
7.2.2  R-CNNs
7.2.3 You Only Look Once
7.2.4 R-CNN Implementation
7.3 Multiple-Target Classification
7.3.1 Preprocessing.

7.3.2 Two-Dimensional Discrete Wavelet Transforms for Noise Reduction
7.3.3 Noisy SAR Imagery Preprocessing by L1-Norm Minimization
7.3.4 Wavelet-Based Preprocessing and Target Detection
7.4 Target Classification
7.5 Multiple-Target Classification: Results and Analysis
7.6 Summary
References
CHAPTER 8 RF Signal Classification
8.1 Introduction
8.2 RF Communications Systems
8.2.1 RF Signals Analysis
8.2.2 RF Analog Signals Modulation
8.2.3 RF Digital Signals Modulation
8.2.4 RF Shift Keying
8.2.5 RF WiFi
8.2.6 RF Signal Detection
8.3 DL-Based RF Signal Classification
8.3.1 DEEP Learning for Communications
8.3.2 DEEP Learning for I/Q systems
8.3.3 DEEP Learning for RF-EO Fusion Systems
8.4 DL Communications Research Discussion
8.5 Summary
References
CHAPTER 9 Radio Frequency ATR Performance Evaluation
9.1 Introduction
9.2 Information Fusion
9.3 Test and Evaluation
9.3.1 Experiment Design
9.3.2 System Development
9.3.3 Systems Analysis
9.4 ATR Performance Evaluation
9.4.1 Confusion Matrix
9.4.2 Object Assessment from Confusion Matrix
9.4.3 Threat Assessment from Confusion Matrix
9.5 Receiver Operating Characteristic Curve
9.5.1 Receiver Operating Characteristic Curve from Confusion Matrix
9.5.2 Precision-Recall from Confusion Matrix
9.5.3 Confusion Matrix Fusion
9.6 Metric Presentation
9.6.1 National Imagery Interpretability Rating Scale
9.6.2 Display of Results
9.7 Conclusions
References
CHAPTER 10 Recent Topics in Machine Learning for Radio Frequency ATR
10.1 Introduction
10.2 Adversarial Machine Learning
10.2.1 AML for SAR ATR
10.2.2 AML for SAR Training
10.3 Transfer Learning
10.4 Energy-Efficient Computing for AI/ML
10.4.1 BM's TrueNorth Neurosynaptic Processor
10.4.2 Energy-Efficient Deep Networks.

10.4.3 MSTAR SAR Image Classification with TrueNorth
10.5 Near-Real-Time Training Algorithms
10.6 Summary
References
About the Authors
Index.