Neural Networks and Deep Learning: A Textbook

by Charu C. Aggarwal

This book covers both classical and modern models in deep learning. The primary focus is on the theory and algorithms of deep learning. The theory and algorithms of neural networks are particularly important for understanding important concepts, so that one can understand the important design concepts of neural architectures in different applications. Why do neural networks work? When do they work better than off-the-shelf machine-learning models? When is depth useful? Why is training neural networks so hard? What are the pitfalls? The book  is also rich in discussing different applications in order to give the practitioner a flavor of how neural architectures are designed for different types of problems. Applications associated with many different areas like recommender systems, machine translation, image captioning, image classification, reinforcement-learning based gaming, and text analytics are covered. The chapters of this book span three The basics of neural  Many traditional machine learning models can be understood as special cases of neural networks.  An emphasis is placed in the first two chapters on understanding the relationship between traditional machine learning and neural networks. Support vector machines, linear/logistic regression, singular value decomposition, matrix factorization, and recommender systems are shown to be special cases of neural networks. These methods are studied together with recent feature engineering methods like word2vec. Fundamentals of neural A detailed discussion of training and regularization is provided in Chapters 3 and 4. Chapters 5 and 6 present radial-basis function (RBF) networks and restricted Boltzmann machines. Advanced topics in neural Chapters 7 and 8 discuss recurrent neural networks and convolutional neural networks. Several advanced topics like deep reinforcement learning, neural Turing machines, Kohonen self-organizing maps, and generative adversarial networks are introduced in Chapters 9 and 10. The book is written for graduate students, researchers, and practitioners.   Numerous exercises are available along with a solution manual to aid in classroom teaching. Where possible, an application-centric view is highlighted in order to provide an understanding of the practical uses of each class of techniques.

Genres: Artificial Intelligence, Nonfiction, Computer Science, Academic, Science, Textbooks, Computers

524 pages, Hardcover

Published September 30, 2018

Reviews

[Raed · Rating 4 out of 5]

this book need a solid understanding of mathematical concepts , and +- all machine learning algorithm, and it's a a good choice for those who want to gain a very strong knowledge in data science engineering

[Filip Karlo Došilović · Rating 4 out of 5]

This is a great book for those who want to gain knowledge about (deep) neural networks. But beware, the book assumes a solid understanding of basic principles and concepts in machine learning.

There are two reasons why I gave the book 4/5:
1. The book seems like a brain dump - a compilation of facts, that the author gained through years of experience, of each model. Do not get me wrong, each model is motivated and presented very well, but the author sometimes just rambles about various facts about the model's behavior - something students usually get as a laboratory exercise.
2. Inconsistent, or maybe nonstandard, notation - the author just does not care about dimensions of vectors and matrices. I did not like this very much.

I like the author's approach to start by representing shallow models (SVM, logistic and linear regression, ...) as a one-layer neural network. I also really liked the extensive treatment of backpropagation through feedforward neural networks and the inclusion of some older models, such as RBF networks.

Regardless of the two cons I mentioned, I would recommend this book to anyone who wants to gain knowledge of deep learning.

[Scott Holstad · Rating 4 out of 5]

Pretty solid work, possibly impressive even. I've not actually delved too deeply into this subject, but from what I have read, I thought this book was definitely among the best I've encountered so far. Recommended.

[Brian Powell · Rating 4 out of 5]

Some good topics and fairly clear explanations but goddam this dude can ramble.

[Max · Rating 2 out of 5]

I read this because I thought it would be nice to review some fundamentals and to get some exposure to topics that I don't work with on a day-to-day basis. It was ok.

The authors rambles a lot. Each topic is covered by a brain dump of more or less everything that came to his mind. Many stretches are super repetitive and needlessly verbose, but vague once things get complicated or technical. It's so obvious that the goal was to fill a five hundred page volume as fast as possible. Many derivations are spotty with loopy arguments. Some chapter clearly haven't ever been proofread at all by anybody. There are so many typos and sentences just missing words. I guess Springer has no editors.

Being from 2018 this is also a bit outdated, of course, but that can hardly be held against the book. There is no mention of Transformers yet (introduced June 2017), but, outside of Machine Translation, people probably didn't immediately see the significance. They grew to such importance in the mean time that any overview text that doesn't cover them needs to be updated to a new edition, presto.

All in all, I think most people would be better of reading Wikipedia articles, blog posts, and actual breakthrough papers.

[Cristián S · Rating 4 out of 5]

It is a great book to learn everything regarding neural networks. It covers all topics needed to master current models used in industry, without the mathematical details that sometimes are a burden in other books. It has a practical approach in terms of understanding the different components of the models, and the most important neural network architectures used in practice. Unfortunately, this field goes so fast that some recent things like transformers are missed, but, it is understandable and probably a following edition will have them and newer things. I didn't like that it is sometimes somewhat disorganized on how figures and text are displayed, and things are repeated many times, showing that the edition of the book didn't go through it completely removing unfinished paragraphs. I would have loved more mathematical treatment of the models, but it is ok to understand the concepts and architectures.

[Şafak Bilici · Rating 5 out of 5]

This book is for those who have studied classical Machine Learning and Statistical Methods, and you want soft introduction to Deep Learning. Good structured comprehensive content, easy language, from intuition to formal definitions. Do not hesitate to choose this book as a starting point for Deep Learning.

[Terran M · Rating 5 out of 5]

This is an excellent 2nd book on deep neural networks; I recommend that you read it after you have finished reading and understanding Goodfellow and perhaps gotten a bit of practice. I would not recommend reading this book cold with no previous experience, as you are likely to find the mathematical derivations for backpropagation and related algorithms too terse to follow as a first exposure. The writing is generally good, other than the fact, that Aggarwal likes commas, too much. In the last chapter, there were some errors in the questions which suggested hurried editing; the rest of the book was solid.

I think Aggarwal is an underrated author whose books deserve more attention, and I have yet to be disappointed by anything of his that I've read.

[Mario Bosnar · Rating 5 out of 5]

A great book for ppl who already worked with ANN and want to remind themselves.