Explainable Machine Learning Models and Architectures

EXPLAINABLE MACHINE LEARNING MODELS AND ARCHITECTURES This cutting-edge new volume covers the hardware architecture implementation, the software implementation approach, and the efficient hardware of machine learning applications. Machine learning and deep learning modules are now an integral part of many smart and automated systems where signal processing is performed at different levels. Signal processing in the form of text, images, or video needs large data computational operations at the desired data rate and accuracy. Large data requires more use of integrated circuit (IC) area with embedded bulk memories that further lead to more IC area. Trade-offs between power consumption, delay and IC area are always a concern of designers and researchers. New hardware architectures and accelerators are needed to explore and experiment with efficient machine-learning models. Many real-time applications like the processing of biomedical data in healthcare, smart transportation, satellite image analysis, and IoT-enabled systems have a lot of scope for improvements in terms of accuracy, speed, computational powers, and overall power consumption. This book deals with the efficient machine and deep learning models that support high-speed processors with reconfigurable architectures like graphic processing units (GPUs) and field programmable gate arrays (FPGAs), or any hybrid system. Whether for the veteran engineer or scientist working in the field or laboratory, or the student or academic, this is a must-have for any library.

Table of Contents:
Preface xiii Acknowledgements xv 1 A Comprehensive Review of Various Machine Learning Techniques 1 Pooja Pathak and Parul Choudhary 1.1 Introduction 1 1.1.1 Random Forest 2 1.1.2 Decision Tree 3 1.1.3 Support Vector Machine 4 1.1.4 Naive Bayes 5 1.1.5 K-Means Clustering 6 1.1.6 Principal Component Analysis 6 1.1.7 Linear Regression 6 1.1.8 Logistic Regression 7 1.1.9 Semi-Supervised Learning 8 1.1.10 Transductive SVM 9 1.1.11 Generative Models 9 1.1.12 Self-Training 9 1.1.13 Relearning 9 1.2 Conclusions 9 2 Artificial Intelligence and Image Recognition Algorithms 11 Siddharth, Anuranjana and Sanmukh Kaur 2.1 Introduction 12 2.2 Traditional Image Recognition Algorithms 13 2.2.1 Harris Corner Detector (1988) 13 2.2.2 SIFT (2004) 15 2.2.3 ASIFT 16 2.2.4 SURF (2006) 17 2.3 Neural Network-Based Algorithms 21 2.4 Convolutional Neural Network Architecture 22 2.5 Various CNN Architectures 23 2.5.1 LeNet-5 (1998) 23 2.5.2 AlexNet (2012) 24 2.5.3 VGGNet (2014) 24 2.5.4 GoogleNet (2015) 24 3 Efficient Architectures and Trade-Offs for FPGA-Based Real-Time Systems 31 L.M.I. Leo Joseph, J. Ajayan, Sandip Bhattacharya and Sreedhar Kollem 3.1 Overview of FPGA-Based Real-Time System 31 3.1.1 Key Elements of Real-Time System 32 3.1.2 Real-Time System and its Computation 32 3.1.3 FPGA Functionality and Applications 33 3.1.4 FPGA Applications 33 3.1.5 FPGA Architecture 34 3.1.6 Reconfigurable Architectures 35 3.2 Hybrid FPGA Configurations and its Algorithms 38 3.2.1 Hybrid FPGA 38 3.2.2 Hybrid FPGA Architecture 39 3.2.3 Hybrid FPGA Configuration 40 3.3 Hybrid FPGA Algorithms 42 3.3.1 Relevance of Hardware-Accelerated Architecture to FPGA Software Implementation 44 3.4 CNN Hardware Accelerator Architecture Overview 46 3.5 Summary 47 4 A Low-Power Audio Processing Using Machine Learning Module on FPGA and Applications 49 Suman Lata Tripathi, Dasari Lakshmi Prasanna and Mufti Mahmud 4.1 Introduction 49 4.2 Existing Machine Learning Modules and Audio Classifiers 50 4.3 Audio Processing Module Using Machine Learning 56 4.4 Application of Proposed FPGA-Based ML Models 57 4.5 Implementation of a Microphone on FPGA 59 4.6 Conclusion 60 4.7 Future Scope 60 5 Synthesis and Time Analysis of FPGA-Based DIT-FFT Module for Efficient VLSI Signal Processing Applications 65 Siba Kumar Panda, Konasagar Achyut and Dhruba Charan Panda 5.1 Introduction 66 5.2 Implementation of DIT-FFT Algorithm 67 5.2.1 A Quick Overview of DIT-FFT 67 5.2.2 Algorithmic Representation with Example 69 5.2.3 Simulated Output Waveform 69 5.3 Synthesis of Designed Circuit 71 5.4 Static Timing Analysis of Designed Circuit 73 5.5 Result and Discussion 77 5.6 Conclusion 77 6 Artificial Intelligence-Based Active Virtual Voice Assistant 81 Swathi Gowroju, G. Mounika, D. Bhavana, Shaik Abdul Latheef and A. Abhilash 6.1 Introduction 82 6.2 Literature Survey 83 6.3 System Functions 87 6.4 Model Training 88 6.5 Discussion 90 6.5.1 Furnishing Movie Recommendations 91 6.5.2 KNN Algorithm Book Recommendation 92 6.6 Results 93 6.7 Conclusion 102 7 Image Forgery Detection: An Approach with Machine Learning 105 Madhusmita Mishra, Silvia Tittotto and Santos Kumar Das 7.1 Introduction 105 7.2 Historical Background 107 7.3 CNN Architecture 109 7.4 Analysis of Error Level of Image 113 7.5 Proposed Model of Image Forgery Detection, Results and Discussion 115 7.6 Conclusion 118 7.7 Future Research Directions 119 8 Applications of Artificial Neural Networks in Optical Performance Monitoring 123 Isra Imtiyaz, Anuranjana, Sanmukh Kaur and Anubhav Gautam 8.1 Introduction 123 8.2 Algorithms Employed for Performance Monitoring 129 8.2.1 Artificial Neural Networks 129 8.2.2 Deep Neural Networks 130 8.2.3 Convolutional Neural Networks 131 8.2.3.1 Convolutional Layer 131 8.2.3.2 Non-Linear Layer 132 8.2.3.3 Pooling Layer 132 8.2.3.4 Fully Connected Layer 132 8.2.4 Support Vector Regression (SVR) 133 8.2.5 Support Vector Machine (SVM) 133 8.2.6 Kernel Ridge Regression (KRR) 133 8.2.7 Long Short-Term Memory (LSTM) 133 8.3 Artificial Intelligence (AI) Methods, Performance Monitoring and Applications in Optical Networks 134 8.3.1 Performance Monitoring 134 8.3.2 Applications of AI in Optical Networking 135 8.4 Optical Impairments and Fault Management 135 8.4.1 Noise 135 8.4.2 Distortion 135 8.4.3 Timing 136 8.4.4 Component Faults 136 8.4.5 Transmission Impairments 137 8.4.6 Fault Management in Optical Network 137 8.5 Conclusion 138 9 Website Development with Django Web Framework 141 Sanmukh Kaur, Anuranjana and Yashasvi Roy 9.1 Introduction 141 9.2 Salient Features of Django 142 9.2.1 Complete 142 9.2.2 Versatile 142 9.2.3 Secure 142 9.2.4 Scalable 143 9.2.5 Maintainable 143 9.2.6 Portable 143 9.3 UI Design 143 9.3.1 HTML 143 9.3.2 CSS 144 9.3.3 Bootstrap 144 9.4 Methodology 144 9.5 UI Design 144 9.6 Backend Development 148 9.6.1 Login Page 148 9.6.2 Registration Page 149 9.6.3 User Tracking 149 9.7 Ouputs 150 9.8 Conclusion 152 10 Revenue Forecasting Using Machine Learning Models 155 Yashasvi Roy and Sanmukh Kaur 10.1 Introduction 155 10.2 Types of Forecasting 156 10.2.1 Qualitative Forecasting 156 10.2.1.1 Industries That Use Qualitative Forecasting 157 10.2.1.2 Qualitative Forecasting Methods 158 10.2.2 Quantitative Forecasting 158 10.2.2.1 Quantitative Forecasting Methods 159 10.2.3 Artificial Intelligence Forecasting 160 10.2.3.1 Artificial Neural Network (ANN) 160 10.2.3.2 Support Vector Machine (SVM) 161 10.3 Types of ML Models Used in Finance 162 10.3.1 Linear Regression 162 10.3.1.1 Simple Linear Regression 162 10.3.1.2 Multiple Linear Regression 162 10.3.2 Ridge Regression 163 10.3.3 Decision Tree 164 10.3.3.1 Prediction of Continuous Variables 164 10.3.3.2 Prediction of Categorical Variables 165 10.3.4 Random Forest Regressor 165 10.3.5 Gradient Boosting Regression 166 10.3.5.1 Advantages of Gradient Boosting 167 10.4 Model Performance 167 10.4.1 R-Squared Method 167 10.4.2 Mean Squared Error (MSE) 167 10.4.3 Root Mean Square Error (RMSE) 168 10.5 Conclusion 168 11 Application of Machine Learning Optimization Techniques in Wind Resource Assessment 171 Udhayakumar K. and Krishnamoorthy R. 11.1 Introduction 172 11.2 Wind Data Analysis Methods 173 11.2.1 Wind Characteristics Parameters 173 11.2.2 Wind Speed Distribution Methods 173 11.2.3 Weibull Method 174 11.2.4 Goodness of Fit 175 11.3 Wind Site and Measurement Details 175 11.3.1 Seasonal Wind Periods 176 11.3.2 Machine Learning and Optimization Techniques 176 11.3.2.1 Moth Flame Optimization (MFO) Method 176 11.4 Results and Discussions 180 11.4.1 Wind Characteristics 182 11.4.1.1 Kayathar Station (Onshore) 182 11.4.1.2 Gulf of Khambhat (Gujarat Offshore) Station 187 11.4.1.3 Jafrabad (Gujarat-Nearshore) 192 11.4.2 Wind Distribution Fitting 195 11.4.2.1 Kayathar Station (Onshore) 196 11.4.2.2 Bimodal Behaviour 196 11.4.2.3 Gulf of Khambhat (Offshore) Wind Distribution 202 11.4.2.4 Jafrabad Station (Nearshore) Distribution Fitting 203 11.4.3 Optimization Methods for Parameter Estimation 212 11.4.3.1 Optimization Parameters Comparison 212 11.4.4 Wind Power Density Analysis (WPD) 214 11.4.4.1 Comparison of Wind Power Density 215 11.5 Research Summary 221 11.6 Conclusions 222 12 IoT to Scale-Up Smart Infrastructure in Indian Cities: A New Paradigm 227 Indu Bala, Simarpreet Kaur, Lavpreet Kaur and Pavan Thimmavajjala 12.1 Introduction 228 12.2 Technological Progress: A Brief History 229 12.3 What is the Internet of Things (IoT)? 230 12.4 Economic Effects of Internet of Things 230 12.5 Infrastructure and Smart Infrastructure: The Difference 232 12.5.1 What is Smart Infrastructure? 233 12.5.2 What are the Principles of Smart Infrastructure? 234 12.5.3 Components of IoT-Based Smart City Project 235 12.6 Architecture for Smart Cities 236 12.6.1 Networking Technologies 237 12.6.2 Network Topologies 237 12.6.3 Network Architectures 238 12.6.3.1 Home Area Networks (HANs) 238 12.6.3.2 Field/Neighborhood Area Networks (FANs/NANs) 238 12.6.3.3 Wide Area Networks (WANs) 238 12.6.3.4 Network Protocols 238 12.7 IoT Technology in India’s Smart Cities: The Current Scenario 239 12.8 Challenges in IoT-Based Smart City Projects 243 12.8.1 Technological Challenges 243 12.8.1.1 Privacy and Security 243 12.8.1.2 Smart Sensors and Infrastructure Essentials 243 12.8.1.3 Networking in IoT Systems 244 12.8.1.4 Big Data Analytics 244 12.8.2 Financial - Economic Challenges 244 12.9 Role of Explainable AI 245 12.10 Conclusion and Future Scope 246 References 246 Index 251