Essentials of Deep Learning and AI: Experience Unsupervised Learning, Autoencoders, Feature Engineering, and Time Series Analysis with TensorFlow, Keras, and scikit-learn

Cover Page
Title Page
Copyright Page
Foreword
Dedication Page
About the Authors
About the Reviewer
Acknowledgement
Preface
Errata
Table of Contents
1. Introduction
    Structure
    Objectives
    1.1 Artificial intelligence
        1.1.1 What is Artificial Intelligence?
        1.1.2 Definitions of Artificial Intelligence
        1.1.3 Applications of Artificial Intelligence
        1.1.4 Industry domains and sectors along with sample use cases
        1.1.5 Broad classification of what is AI, ML, FL, and DL?
    1.2 Machine learning
        1.2.1 History and definition of ML
        1.2.2 Machine learning and its applications
        1.2.3 Classification of ML algorithms
    1.3 Deep Learning
        1.3.1 What are the prerequisites to understand deep learning?
        1.3.2 Difference between machine learning and deep learning
        1.3.3 Applications of deep learning
    1.4 Tools and frameworks for AI, ML and DL
    1.5 Languages used for AI, ML, and DL
    1.6 Sample datasets for AI, ML, and DL development
    Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
2. Supervised Machine Learning
    Structure
    Objectives
    2.1 Introduction to Supervised Machine Learning
    2.2 Data Cleanup
    2.3 Data preparation
    2.4 Classification and regression
    2.5 Architecture and realization of algorithms
        2.5.1 Linear Regression
        2.5.2 Support vector machine (SVM)
        2.5.3 Decision trees
        2.5.4 Random forest
    2.6 Performance statistics
        2.6.1 Performance metrics of regression problems
        2.6.2 Performance metrics of classification problems
    2.7 Optimization and loss methods
    2.8 Use cases and examples
    2.9 Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
3. System Analysis with Machine Learning/Un-Supervised Learning
    Structure
    Objectives
    3.1 Introduction and architecture of unsupervised machine learning
    3.2 Data preparation methods and steps
        3.2.2 Data preprocessing and scaling
            3.2.2.1 StandardScaler
            Code sample
            3.2.2.2 MinMaxScaler
            3.2.2.3 RobustScaler
            3.2.2.4 Normalizer
    3.3 Clustering techniques
        3.3.1 K-Means
        3.3.2 Hierarchical clustering
        3.3.3 Density-based spatial clustering of applications (DBSCAN)
    3.4 Other algorithms and methodologies
    (Dimensionality reduction techniques)
        3.4.1 t-Distributed stochastic neighbor embedding (t-SNE)
        3.4.2 Principal Component Analysis (PCA)
            3.4.2.1 Decomposition incremental PCA
            3.4.2.2 Decomposition kernel PCA
            3.4.2.3 Decomposition MiniBatchSparse PCA
            3.4.2.4 Decomposition PCA
            3.4.2.5 Decomposition sparse PCA
        3.4.3 Singular value decomposition (SVD)
        3.4.4 Independent component analysis (ICA)
        3.4.5 Dictionary learning
    3.5 Error minimization
        3.5.1 Distance computing
        3.5.2 Threshold limit
        3.5.3 Log loss
        3.5.4 Euclidian distance
        3.5.5 Examples and samples
    3.9 Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
4. Feature Engineering
    Why is feature engineering needed?
    Structure
    Objectives
    4.1 Introducing feature engineering
    4.2 What is Feature selection?
        4.2.1 Baselining model
        4.2.2 Categorical encodings
        Nominal encoding types
            4.2.2.1 One Hot Encoding
            4.2.2.2 Mean encoding
            4.2.2.3 One hot encoding with many categories
        Ordinal encoding types
            4.2.2.4 Label encoding
            4.2.2.5 Target guided ordinal encoding
            4.2.2.6 Frequency encoding
            4.2.2.7 Other Encoding techniques
            Hash encoder
            Effect encoding
            Dummy encoding
            Binary encoding
        4.2.3 Feature generation
        4.2.4 Feature Selection
            Correlation
            Feature importance methods
            Univariate feature selection
        4.2.5 Collecting and refining data
            4.2.6 Cleaning and organizing data
        4.2.7 Data preparation
        4.2.8 Mining data for pattern selection
    4.3 Other popular techniques of feature engineering
        4.3.1 Imputation
        4.3.2 Handling outliers
        4.3.3 Binning
        4.3.4 One-hot encoding
        4.3.5 Feature split
        4.3.6 Scaling
    4.4 Examples and samples
    Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
5. Classification, Clustering, Association Rules, and Regression
    Structure
    Objectives
    5.1 Introduction
    5.2 Classification Techniques
    5.3 One-Class Classification
    5.4 Zero-Shot Learning
    5.5 One Shot, Few Shot, K-Shot, or N-Shot Learning
    5.6 Clustering techniques
    5.7 Distribution-Based Clustering
        5.7.1 Density-Based Clustering
        5.7.2 Fuzzy Clustering
        5.7.3 Grid-Based Clustering
    5.8 Association Rules Techniques
    5.9 Regression Techniques
    5.10 Logistic Regression
    5.11 Ridge Regression
    5.12 Lasso Regression
    5.13 ElasticNet Regression
    5.14 Factors for selecting the right regression model
    5.15 Use Cases and Examples
    Conclusion
    Points to remember
    Questions
    Multiple Choice Questions
        Answers
6. Time Series Analysis
    Structure
    Objectives
    6.1 Introduction to time series
    6.2 Various types of time series
    6.3 Univariate and multivariate time series models
    6.4 Time domain and frequency domain time series models
    6.5 Linear and non-linear time series models
    6.6 Time series models
        6.6.1 Autoregression (AR)
        6.6.2 Moving average (MA)
        6.6.3 Autoregression Moving Average (ARMA)
        6.6.4 Autoregressive Integrated Moving Average (ARIMA)
        6.6.5 Vector Autoregression (VAR)
        6.6.6 Vector Autoregression Moving Average (VARMA)
        6.6.7 Fourier Transforms (FT)
    6.7 Examples and samples
    Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
7. Data Cleanup, Characteristics and Feature Selection
    Structure
    Objectives
    7.1 Introduction
    7.2 Data formatting
    7.3 Normalization
        Min-Max normalization
        Z-Score Normalization or Standardization
        Box-Cox Transformation
        Decimal Scaling
    7.4 Model Training and Test Splitting
    7.5 Bias and Variance Trade-off
    7.6 Model Overfitting and Underfitting
    7.7 Cross Validation
    7.8 Feature Reduction Techniques
    7.9 Use Cases and Examples
    Conclusion
    Points to remember
    Questions
    Multiple Choice Questions
        Answers
8. Ensemble Model Development
    Structure
    Objectives
    8.1 Introduction
    8.2 Ensemble methods
        8.2.1 Popular ensemble methods
            8.2.1.1 Sequential ensemble methods
            8.2.1.2 Parallel ensemble methods
            8.2.1.3 Averaging ensemble methods
            8.2.1.4 Boosting ensemble methods
    8.3 Combining weak learners
        How to combine weak learners?
    8.4 Advanced ensemble model building tips
    8.5 Hyperparameter tuning
    8.6 Genetic algorithm-based tuning
    8.7 Use cases or examples
    Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
9. Design with Deep Learning
    Structure
    Objectives
    9.1 Introduction
    9.2 Architecture of CNN
        9.3.1 AlexNet
    9.4 Training a CNN network
    9.5 Latest trends and algorithms in CNN
    9.6 Use cases or examples
    Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
10. Design with Multi Layered Perceptron (MLP)
    Structure
    Objectives
    10.1 Introduction
    10.2 Components of MLP
    10.3 Architecture of MLP
    10.4 Training mechanisms in MLP
    10.5 Latest trends in MLP
        10.5.1 Knowledge distillation
    10.6 Use cases or examples on how to build an MLP with various frameworks
    Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
11. Long Short Term Memory Networks
    Structure
    Objectives
    11.1 Introduction
    11.2 Recurrent neural networks (RNN)
    11.3 Gated recurrent units (GRU)
    11.4 Architectures of Long Short Term Memory Networks (LSTM)
        11.4.1 Bi-directional LSTM
        11.4.2 Attention-based LSTM
    11.5 Training mechanisms of LSTM
    11.6 Use cases and examples
    11.7 NLP using LSTM and advancements
    Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
12. Autoencoders
    Structure
    Objectives
    12.1 Introduction to autoencoders and simple architecture
    12.2 Undercomplete autoencoders
    12.3 Overcomplete autoencoders
    12.4 Denoising autoencoders
    12.5 Sparse autoencoders
    12.6 Stacked autoencoders
    12.7 Variational autoencoders (VAEs)
    12.8 Other autoencoders
    12.9 Examples and samples
    Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
13. Applications of Machine Learning and Deep Learning
    Structure
    Objectives
    13.1 Introduction
    13.2 Domain-specific applications
        13.2.1 Telecommunications
    13.3 Technology specific applications
        13.3.1. The working mechanism
    13.4 Device specific applications
        13.4.1 Feedback predictor model
    13.5 Platform-specific applications
    13.6 Solution-specific application
        13.6.1 Contextual advice
        13.6.2. Context-based search
    Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
14. Emerging and Future Technologies
    Structure
    Objectives
    14.1 Introduction to next-generation technologies
        14.2 Internet of Things
        14.2.1 IoT issues
    14.3 Cloud computing
    14.4 5G Networks
        14.4.1 Resource management in 5G
        14.4.2 Prediction based architecture
        14.4.3 Resource allocation
    14.5 Quantum computers
    14.6 Conversation systems
    14.7 Neuromorphic computing
    14.8 Deep reinforcement learning
    Conclusion
    Points to remember
    Questions
    Multiple choice questions
        Answers
Index