Advanced Deep Learning with TensorFlow 2 and Keras: Apply DL, GANs, VAEs, deep RL, unsupervised learning, object detection and segmentation, and more, 2nd Edition

This book gives a very quick introduction to the core ML models (MLP, CNN, RNN), and I was looking forward to learning about advanced techniques for each of them. I'm an NLP engineer, and so was rather disappointed to see that the rest of the book is aimed almost entirely at computer vision techniques, with some reinforcement learning towards the end. Sadly this book provided me with nothing new in terms of natural language processing techniques.Also the explanations aren't very detailed, so you will probably need to purchase a companion book to go alongside this, as it's mostly discussed from a practical application with lots of assumed knowledge (even about the more advanced topics).Finally, the written form/language coherence is slightly off, which isn't helpful when trying to get your head around the very brief explanations of advanced concepts.

That's really what I needed. A very well written book an advanced deep learning techniques. I have mostly focused my attention on autoencoders and Gans, but the other chapters sound interesting either.

This book is a comprehensive guide to Deep Learning with TensorFlow/Keras.The first chapter introduces three varieties of Deep Learning Networks: Multi-Layer Perceptrons (MLP), Convolutional Neural Networkss (CNN) and Recurrent Neural Networks (RNNS) and discusses their differences and relative strengths: MLPs are commonly used in logistic/linear regression problems; RNNs are sequential popular for sequential data input. and CNNs excel in problems involving multi-dimensional data like images and videos.THe use of each type of neural network is illustrated by applying them tothe MNIST dataset of handwritten digits, with the MLP example script achievinga test accuracy of 98 %, the CNN example achieving an accuracy of 98.7 %, andthe RNN example achieving an accuracy of 98 %.Chapter 2 discusses Deep Neural Networks, focusing on the network types ResNet(which introduced the concept of Residual Learning) and DenseNet (which representsan improvement of TesNet).Further ino the book, Chapter 4 discusses GANs, Generative Adversial Networks, used(for instance) to create synthetic images. Chapter 5 then discusses Improved GANsChapter 7 discuses cross-domain transfer (translating an image from one form toanother, such as transforming satellite images to maps, or changing the artworkstyle of one artist to another) and discusses the application of GANs to thesetasks.Some other topics of discussion:Deep Reinforcement Learning (Chapter 8), Object Detection (Chapter 11) andUnsupervised Learning (Chapter 13, introducing the concept of Mutual Information).All in all a thorough and useful introduction to TensorFLow 2/Keras.

I received my copy directly from the publisher, but I wish to share my thoughts on Amazon. I find this book to be very interesting – the author is highly knowledgeable in this field and has a number of interesting publications. This book covers the exciting Keras deep learning library and why it is so capable in this brilliant field. My favourite aspect is Deep Neural Networks, which have shown excellent performance in their accuracy of their classification. I would add that when approaching this book, it would help the reader to brush up on mathematical ideas and concepts like calculus, linear algebra, stats and probability at university level. There are lots mathematical ideas which would benefit from familiarity and that will help the reader to better understand the concepts and to appreciate the book. The code examples are very thorough and detailed with comments about what is happening in the Python code. There are many good links to arXiv articles which are relevant to the discussion and the source code for the book is also available on GitHub. I can highly recommend this book!