Deep Belief Nets in C++ and CUDA C: Volume II: Autoencoding in the Complex Domain

As its name implies, this book is better for practitioners, not researchers, given its light treatment of the theories and heavy emphasis on the actual code. As a researcher I personally do not think I learned much from this book, but there is no doubt that at some point I will find it useful as a good codebook reference.

Still working my way through this, trying to learn CUDA and OpenCL. It is a good example of an application of neural network application.

Everything went very smooth. Love the book!

Having read Volume 1 and appreciated the amount of attention paid to implementation, I was eager to start on Volume II. This Volume is well worth it! He explains so much information in a very clear manner, that most theoretical books completely ignore. This rare information is a must have/must read, and you have to check out his software made available for free, DEEP and CONVNET (Volume 3), along with TSSB. You won't be dissapointed.

This is a good resource for learning Deep Belief Nets, especially considering the limited resources available online on this topic.

First, as I stated in my review of Volume 1, I must disclose that I have known Dr. Masters for 20 years and have collaborated with him on various projects including a book we co-authored.(Statistically Sound Machine Learning for Algorithmic Trading of Financial Instruments). In addition he was a crucial adviser on my book Evidence Based Technical Analysis. He is also a friend.Amazon permits you to view the contents of the book so I won't take up space doing that. Suffice it to say that if you purchased Volume I and found it as important as I did - Deep Belief (Learning) Networks are the most important advance in machine learning in the last decade or two - then Volume II is required reading.Tim Masters has a writing style that never talks down to the reader yet it offers a lot for the uninitiated. At the same time it has the requisite information for those already involved with machine learning.Part of me wishes these two volumes had not been written because I think he gives away too much for too little. That would not have been my choice but it was the authors'. The discussion of time series problems, especially the notion that deep nets can utilize information found in the trajectory of the feature vector through feature space was of particular interest.The free software is user friendly and useful.David Aronson

This book is an amazing addition to the first volume. If you are serious about machine learning and artificial intelligence then this is something that you must purchase. It will help you grasp a deep understanding of auto-encoders and machine learning in general so that you may make more informed architectural decisions with your programming models and/or business decisions.

Have purchased many of the author's previous volumes and thought would be nice to add to collection of books and gain some new insights into old techniques being presented. Sadly this volume lacks many of the previous insights previously presented by the author. The splitting of the two volumes frankly could have been bettered handled by incorporating all into just one volume. The new material frankly is disappointing and the accompanying software is of little value and really just a demo program. Better presentations elsewhere and too bad as author had made very significant contributions to this field of study during it's infancy.

Die Hälfte des Buches ist eine abgedruckte Bedienungsanleitung.Der Autor bietet ein Programm zum Download an, so weit so gut ABERdie Bedienungsanleitung nimmt ein grossen Teil des dünnen Buches weg.Das Schlimmste, im zweiten Teil macht er das genau so.Der restliche Inhalt des Buches ist unübersichtlich und inhaltlich dünn.Im Internet wird das Thema besser und gratis behandelt.Empfehlung, lasst es besser weg!