Название: Applying Data Science: How to Create Value with Artificial Intelligence Автор: Arthur K. Kordon
Издательство: Springer
Год: 2020
Страниц: 511
Язык: английский
Формат: pdf (true)
Размер: 17.4 MB
This book offers practical guidelines on creating value from the application of data science based on selected artificial intelligence methods. In Part I, the author introduces a problem-driven approach to implementing AI-based data science and offers practical explanations of key technologies: machine learning, deep learning, decision trees and random forests, evolutionary computation, swarm intelligence, and intelligent agents. In Part II, he describes the main steps in creating AI-based data science solutions for business problems, including problem knowledge acquisition, data preparation, data analysis, model development, and model deployment lifecycle. Finally, in Part III the author illustrates the power of AI-based data science with successful applications in manufacturing and business. He also shows how to introduce this technology in a business setting and guides the reader on how to build the appropriate infrastructure and develop the required skillsets.
There is a growing need for a book that demonstrates the value of using AI-driven technologies within the Data Science framework, explains clearly the main principles of the different approaches with less of a focus on theoretical details, offers a methodology for developing and deploying practical solutions, and suggests a roadmap for how to introduce these technologies into a business.
An artificial neural network consists of a collection of processing elements, organized in a network structure. The output of one processing element can be the input to another processing element. Numerous ways to connect artificial neurons into different network architectures exist. The most popular and widely applied neural network structure is the multilayer perceptron. It consists of three types of layer (input, hidden, and output). Neural networks have the capability to learn how to perform certain tasks and adapt themselves by changing the network parameters in a surrounding environment. The requirements for successful adaptive learning are as follows: choosing an appropriate architecture, selecting an effective learning algorithm, and supporting model building with representative training, validation, and test data sets.
Recently another more complex type of neural network, called deep learning networks, has been gaining momentum in academic research as well as in potential business applications. The fundamental difference between classical machine learning and deep learning is that the latter learns how to learn. The key principles of deep learning, the key structures of deep learning networks, and the benefits of this approach are discussed briefly.
The book is ideal for data scientists who will implement the proposed methodology and techniques in their projects. It is also intended to help business leaders and entrepreneurs who want to create competitive advantage by using AI-based data science, as well as academics and students looking for an industrial view of this discipline.
Скачать Applying Data Science: How to Create Value with Artificial Intelligence