Sinkakeisan to shinsogakushu (Evolutionary computation and deep learning - Emergence of intelligence)
192 pages, A5 format
It has been said that Artificial Intelligence (AI) is in its third resurgence in the recent years. One of the reasons may be the development of neural networks known as deep learning. This book provides a detailed explanation of deep learning from the theoretical background of neural networks, which are its foundation, to its recent progress, achievements, and issues.
I specialize in evolutionary computation and emergence, but I have always been interested in neural networks since my undergraduate days and have been using them frequently for research. Evolutionary computation is often used together with neural networks, and it is not uncommon for both themes to be discussed at international academic conferences. As such, this book also discusses neuro-evolution, which has been attracting attention as a method that integrates evolutionary computation and neural networks and has been applied in various fields.
In addition, I personally or indirectly experienced the first and second AI winters from the emergence of the perceptron and neural networks, up to deep learning. Therefore, I consider it meaningful to speak about the history of their rise and fall as well as their background in my own words for the benefit of those who have no knowledge of their boom in the past.
I am secretly impressed that recent students are using deep learning and AI methods to solve problems and think from new perspectives. On the other hand, it is very disappointing that many of them are unaware of the classic masterpieces and historical background of AI and neural networks. Hence, in order to highlight their significance, this book also explains the controversy over symbolism and the cognitive scientific debate. In particular, the final chapter, which is entitled “Emergence of Intelligence,” explains the realization of AI and the evolution of the brain. Since essay presentation was allowed in the field of AI research, this book also includes my own way of thinking to a substantial degree.
With this background, this is an introductory book for beginners on AI. It also explains the approach to the realization of AI and the emergence of intelligence using “evolution” and “learning” as keywords. The book also provides an easy explanation from the basics of the neural network up to deep learning based on theory for the benefit of beginners who have no presupposed knowledge of algorithms or biology. In developing deep learning, the book also touches on the basics of neural networks, which is its predecessor, and the recent research, as well as representative methods and application examples of deep learning. This book also keeps the mathematics to a minimum; as much as possible, mathematical formulas and programs were generally not used.
(Written by IBA Hitoshi, Professor, Graduate School of Information Science and Technology / 2020)
Table of Contents
Chapter 2 Neural networks and learning
Chapter 3 Deep learning
Chapter 4 Evolving neural networks
Chapter 5 Emergence of intelligence
The Okawa Publications Prize (The Okawa Foundation for Information and Telecommunications 2016)
Software and demonstration downloadable from the following link: