Название: Programming Neural Networks with Python: Your Practical Guide to Building Smart AI Systems with Machine Learning
Автор: Joachim Steinwendner, Roland Schwaiger
Издательство: Rheinwerk Publishing
Год: 2025
Страниц: 457
Язык: английский
Формат: epub (true)
Размер: 20.1 MB

Neural networks are at the heart of AI—so ensure you’re on the cutting edge with this guide! For true beginners, get a crash course in Python and the mathematical concepts you’ll need to understand and create neural networks. Or jump right into programming your first neural network, from implementing the scikit-learn library to using the perceptron learning algorithm. Learn how to train your neural network, measure errors, make use of transfer learning, implementing the CRISP-DM model, and more. Whether you’re interested in Machine Learning, gen AI, LLMs, Deep Learning, or all of the above, this is the AI book you need!

This is a hands-on book for getting started with ANNs. Supported by graphics, sample applications, and Python code, we’ll demonstrate how ANNs work and how you can program them yourself. This book consists of two parts. The first part gives you a compact overview from the concept to the implementation of various neural networks. We explain to you in a practice-oriented and clear way what you need to do to build a neural network. You’ll learn about different neural approaches as well. By the end of the first part, you will have programmed and understood multiple functioning networks. You’ll also learn you how to find solutions that work in real-life scenarios.

The second part provides background and context for you because you now want to take a broader approach to the topic and add some theory to your first steps. We discuss the historical development, biological comparisons, different types of networks, and finally different learning algorithms for networks. You’ll get to know more technical terms, learn about more advanced concepts, and be able to classify the individual steps to the finished network into a larger system. Specifically, the Neocognitron network by Kunihiko Fukushima, a pioneer in the field of ANN research, is mentioned here, which in our view was the pioneer for modern convolutional neural networks (CNNs) and ought to receive the appropriate recognition.

To make it as easy as possible for you to get started, we’ve reduced the necessary prior knowledge to a minimum. In two crash courses in the appendixes of the book, we explain important basic principles of Python (Appendix A) and mathematics (Appendix B), and you get a short introduction to TensorFlow version 2 and Keras (Appendix C). We explain all other basics as simply and comprehensibly as possible directly in the practice-oriented chapters. If you have little experience with programming, you should definitely work through Chapter 2 and Appendix B and, if necessary, refer to Python books or tutorials. Rheinwerk Publishing offers fantastic books on this topic. If you already do have programming experience, you can skim through Appendix A to get to know the Python syntax better. The same applies to math skills and Appendix B.

Highlights include:

1) Network creation
2) Network training
3) Supervised and unsupervised learning
4) Reinforcement learning
5) Algorithms
6) Multi-layer networks
7) Deep neural networks
8) Back propagation
9) Transformers
10) Python
11) Mathematical concepts
12) TensorFlow