Introduction to deep training. Chernyak E.

The author of the book is an artificial intelligence researcher specializing in the processing of a natural language, the revolution in which has made deep learning. Unfortunately, it took him a lot of time to understand this. We can say that neural networks threaten the revolution for the third time, and by no means the first. Nevertheless, the author suddenly turned out to be far behind and struggled to catch up. That is why he did what any respecting professor would do in his place: he planned to teach the course and began to expedite the material by looking at the web page. This explains several outstanding features of this book. First, brevity. Secondly, it is highly dependent on the project. The author believes that it is better to study the material in computer science when writing programs, so the book largely reflects its habits in teaching. This book, first of all, is conceived as a textbook for a course for deep learning. The course that the author teaches in Brown is intended for both graduates and other students, and covers all the material. Although the actual amount of material on linear algebra is not so large, the students said that without it it would be quite difficult for them to understand multilayer networks and the tensors they needed. And finally, there is a prerequisite for probability and statistics. The author also implies the elementary knowledge of readers in programming in the Python language. Although this material is not included in the book, the author has an additional “laboratory” on the basics of Python.