Neural networks are among the most aesthetically pleasing programming paradigms ever developed. The traditional programming approach involves providing the computer with instructions by decomposing complex problems into numerous smaller, well-defined tasks that can be executed effortlessly by the computer. In contrast, we do not instruct the computer on how to solve our problem within a neural network. Conversely, it acquires knowledge from empirical data and devises its own resolution to the given dilemma. Learning automatically from data appears promising.
However, with the exception of a few specialized problems, we did not know how to train neural networks to outperform more conventional approaches until 2006. The year 2006 marked a turning point with the identification of methods for acquiring knowledge using deep neural networks. The current term for these methods is deep learning. Deep neural networks and deep learning have undergone significant advancements and now demonstrate exceptional performance across a wide range of critical issues in computer vision, speech recognition, and natural language processing. Major corporations including Google, Microsoft, and Facebook are implementing them extensively.
This book aims to assist readers in mastering the fundamental concepts of neural networks, including contemporary deep learning techniques. Upon completing the book, readers will have developed code that solves complex pattern recognition problems using deep learning and neural networks. And they will have the knowledge and skills necessary to apply neural networks and deep learning to solve problems of their own design.