Artificial Intelligence By Example - Second Edition

This is one of the best authored books on ai.The explanation flows well making it easy to understand.

Too many books are only math oriented and miss the implementation of using such algorithms in programming languages. Other books focus on programming the concepts but don't go in depth on the math behind it. From start to finish this book gives you both worlds. No longer do i have this black mystery box mentality of implementing an algorithm(using a library) knowing how to use it but have no idea how it works behind the scenes. This book combines a mathematician, developer, and has Business logic. I Would highly recommend this book.

I was very interested in looking at this book for the last two chapters - Neuromorphic Computing (NC) and Quantum Computing (QC). I knew nothing about NC, and I have been studying QC. I have some background with machine learning. I found these two chapters to be very fun, informative and easy to read.From the Neuromorphic Computing chapter I learned that Intel Research is working on a neurocomputing chip, containing hundreds to thousands of neurons, and that those chips could be connected to make a network. Given that there are 100 billion neurons in a human brain, it would take a huge number of chips to be equivalent. Neurocomputing software, Nengo, is mentioned, which I found interesting, also in the Further Reading section a book written by the author of Nengo is mentioned which describes "how to build a brain". In the chapter, a distinction is made between the neuromorphic computing approach and neural nets. It is also mentioned that Google's TPU is a specialized chip, that is, hardware designed to work well with the TensorFlow software, and that we can expect more of this in the future.The Quantum Computing chapter is brief, but nice in that it gives a quick introduction, and connects with the previous chapter, Neuromorphic Computing. It is mentioned that in the future, a quantum computer could represent a brain, which could be called from a classical computer. A nice, quick, back-of-the-napkin like calculation is done to demonstrate the difference between linear (classical) and exponential (quantum) growth. The author gives an example of a quantum algorithm he wrote which processes some data and seems to return a number which could be interpreted as a movie "recommended" or not. There's not enough description, I think, of the gates applied to get a good feel about what was done and why, but it's nice in that it's really "to the point". Quantum algorithms seem like a pretty hard topic, but this example is motivating, causing me to think - hey, that sounds neat, it's different from what I had been thinking, what is quantum ML like? and, to seek out other sources on the topic.Finally, at the end of the book there is a section - Answers to the Questions, which were asked at the end of each chapter. I enjoyed reading through these to test myself and check the answers against my thoughts. I liked reading the sections about: 1) the self-driving vehicles, which included a challenge question - would you like to design an autonomous driving system for your city? 2) adding emotional intelligence to chat bots 3) combining methods, for example, reinforcement learning + deep learning, decision trees + k-means clustering, genetic algorithms + neural nets.The author is very bullish on quantum computers, though there is no guarantee that QC's can be developed to the point of being functional, for a real problem. I too, however, am optimistic. Via the chapter questions and answers, the author does comment on what existing systems cannot do today; for example, that there currently is no "general AI", like a human, but instead, just "narrow AI", as in, specific tasks only.I like the breadth.

Disclaimer: The publisher and asked me to review this book and gave me a review copy. I promise to be 100% honest in how I feel about this book, both the good and the less so.Overview:I have looked through this book, and I have to say that I am disappointed. I wasn't quite sure what to expect from the title. Was it a beginner book? A cookbook? A book for experienced practitioners? After reviewing it, I still don't know.What I Like:This book has something that I hadn't seen in other Packt AI title: answers to the chapter exercises. That gave me an initial good feeling about the book, as did seeing that each chapter had a section for Further Reading, something that the more recent Packt titles appear to be doing.Maybe the best thing I have found in this book is how each chapter is structured internally, the microstructure. The author organizes each chapter the same: textual comprehension through a use case, mathematical comprehension, example code, a summary, exercises, and further reading. I personally feel that this is a very structure to have.What I Don't Like:When looking over the TOC, I did not see a well organized structure, a macrostructure. The author starts with reinforcement learning, then talks about datasets, later talks about blockchain, further on the Internet of Things, a few chapters on chatbots, genetic algorithms, and finishes with quantum computing. In between some of these are chapters on CNNs, K-Means clustering, and a few topics. There is no natural flow and some of the topic choices, while interesting, don't really fit.In chapter one, before getting into reinforcement learning, the author spends on section discussing what AI is, followed by his personal learning philosophy. To quote a section on 'Overcoming real-life issues using the three-step approach', "First, begin by understanding the subject as a subject matter expert. Then, write the analysis with words and mathematics to make sure your reasoning reflects the subject and, most of all, that the program will make sense in real life. Finally, in step 3, only write the code when you are sure about the whole project." In an agile world, this kind of programming gets left in the dust.There are other bits and pieces within the book that I scratched my head over, such as suggesting an online encyclopedia article on the life of the inventor of a technique as further reading. One of the exercises, which are typically yes/no questions, is 'Can a human beat a chess engine? (Yes | No)'. And there are odd comments thrown in here and there, such as saying that anyone with a driver's license is an expert driver and so can use driving as an example for problems without consulting anyone else.What I Would Like to SeeMore than anything else, I would like to see a book that is consistently about AI with a macrostructure that flows. To choose the right macrostructure, this book needs to decide what it is. If it's a reference, similar topics need to be within a group of chapters. If it's a beginner's learning book, then it needs to start with a simple topic and work its way toward the more complicated topics. I loved the proposed structure within the chapters (which unfortunately wasn't always followed clearly enough), but ingredients cooked perfectly don't make a god meal if they don't go together.Overall, I give this book a 2.7 out of 5 stars, rounded up to 3. The author is clearly knowledgeable on the subject of AI, but they had trouble communicating it to their audience. I know that I will use this text to gain some important knowledge, but they are unfortunately the diamonds in the rough.

information are helpful , good for students