Machine Learning Theory to Applications: Theory to Applications

Description

The book reviews core concepts of machine learning (ML) while focusing on modern applications. It is aimed at those who want to advance their understanding of ML by providing technical and practical insights. It does not use complicated mathematics to explain how to benefit from ML algorithms. Unlike the existing literature, this work provides the core concepts with emphasis on fresh ideas and real application scenarios. It starts with the basic concepts of ML and extends the concepts to the different deep learning algorithms. The book provides an introduction and main elements of evaluation tools with Python and walks you through the recent applications of ML in self-driving cars, cognitive decision making, communication networks, security, and signal processing. The concept of generative networks is also presented and focuses on GANs as a tool to improve the performance of existing algorithms. In summary, this book provides a comprehensive technological path from fundamental theories to the categorization of existing algorithms, covers state-of-the-art, practical evaluation tools and methods to empower you to use synthetic data to improve the performance of applications. Cover Title Page Copyright Page Preface Table of Contents Abbreviations 1. Introduction 2. Linear Algebra Matrix rules Eigenvalues and eigenvectors LU decomposition Statistics and probabilities Momentums Expectation Multivariate distributions Cauchy distribution Dirichlet distribution Multimodal distribution Student’s t distribution Gaussian distribution 3. Machine Learning Machine learning approaches Historical background Data mining Optimization Statistics Theory Different kinds of learning algorithms Supervised learning Unsupervised learning K means clustering Principal component analysis Semi-supervised learning Reinforcement learning Self-learning Feature learning Machine learning models Linear regression Logistic regression K nearest neighbor classifier Naïve Bayesian classifier Artificial neural networks Decision trees and random forests Support vector machines Bayesian networks Genetic algorithms Federated learning 4. Some Practical Notes Resampling method Cross validation Leave one out cross validation K-fold cross validation Metrics Accuracy Precision Recall F1 score Normalization Overfitting and underfitting Regularization Ridge regression Lasso regression Dropout regularization Ceiling analysis 5. Deep Learning Overview Interpretations Artificial neural networks Deep neural networks Deep learning algorithms Convolutional neural networks Recurrent neural networks Long short term memory networks Generative adversarial networks Radial basis function networks (RBFNs) Multi-layer perceptrons (MLP) Deep belief networks Restricted Boltzmann machines Autoencoders Challenges 6. Generative Adversarial Networks Generative Adversarial Networks (GANs) Conditional GAN (CGAN) Auxiliary Classifier GAN (AC-GAN) Wasserstein GAN (WGAN) WGAN with Gradient Penalty (WGAN-GP) Info GAN Least Square GAN (LSGAN) Bidirectional GAN (BiGAN) Dual GAN Deep Convolutional GAN (DCGAN) 7. Implementation Accelerated computing Machine learning frameworks and libraries No need for special hardware support Interactive data analytic and visualization tools Deep learning frameworks and libraries TensorFlow Keras Microsoft CNTK Caffe Caffe2 Torch PyTorch MXNet Chainer Theano Deep learning wrapper libraries References Index