Characteristics of Neural Networks based on Energy Functions. Characteristics and Mechanical Properties of Neural Networks based on Energy Functions

      Аннотация: This book, separated in five chapters is about the Characteristics and Mechanical Properties of Artificial Neural Networks based on Energy Functions. In the first chapter, the author gives the background, purpose and outline of the rest of the book. Chapter 2 titled "Neural Networks based on Energy Functions", describes the continuous Hopfield Model and the Inverse Function Delayed Model. It then solves Combinatorial Optimization Problems using these types of networks. This provides most of the basic knowledge needed to read the rest of the book. Chapter 3 discusses the Tau U=0 model characteristics including the update methods, stability of the solution and the discrete tau u=0 analysis which is one of the main concerns of this book. Chapter 4 introduces a new algorithm for solving combinatorial optimization problems which is based on force analysis. Finally in Chapter 5 the conclusions are given. This work was done as part of my work in order to graduate in Tohoku University, Japan. It was done whilst doing research at the Laboratory for Brainware - Nanoelectronics and Spintronics Research Institute.

Ключевые слова: Artificial Intelligence, neural networks, Hopfield, Energy Functions, Combinatorial Optimizarion Problems, Inverse Function Delayed Model