Computational Methods for Deep Learning: Theoretic, Practice and Applications

Integrating concepts from deep learning, machine learning, and artificial neural networks, Computational Methods for Deep Learning: Theoretic, Practice and Applications presents content progressively from easy to more complex, orienting its content about knowledge transfer from the viewpoint of machine intelligence. It adopts the methodology from graphical theory, mathematical models, and algorithmic implementation, as well as covers datasets preparation, programming, results analysis and evaluations.

Beginning with a grounding about artificial neural networks with neurons and the activation functions, the work then explains the mechanism of deep learning using advanced mathematics. In particular, it emphasizes how to use the latest MATLAB deep-learning toolboxes for implementing deep learning algorithms.

As a prerequisite, readers should have a solid understanding especially of mathematical analysis, linear algebra, numerical analysis, optimizations, differential geometry, manifold, and information theory, as well as basic algebra, functional analysis, and graphical models.

This textbook is aimed at Computer Science research students and engineers, as well as scientists interested in deep learning for theoretic research and analysis. This book is also helpful for those researchers who are interested in machine intelligence, pattern analysis, natural language processing, and machine vision.

The book uses MATLAB throughout and also includes examples of the Deep Learning Toolbox and Image Processing Toolbox.