Machine Learning: An Applied Mathematics Introduction

Coming from a non mathematical backround the explanation of each algorithm and idea presented in the book was very easy to grasp (although I got lost when trying to follow the more advanced equations). This book has really improved my understanding on the topic and I am giving it a second read to fully understand all the math. This book is a good choice for a layman who wants to dive head first into the topic and doesn't mind having some of the mathematical principles goes over their head the first run through.

Great resource. It’s like talking to someone one who is just giving you the simple straight answer to what’s going on. This book’s tone and depth is between the buzz word laden “intro to machine learning” books for business people and the “too much math for non majors” textbooks that focus a specific type of machine learning. With that said I use it to gain an intuition and the first layers of mathematical depth to each ML algorithm. I believe that this does not replace a textbook but more of a straightforward companion. Highly recommend.

This is an informal introduction to machine learning techniques and philosophy. It is an easy reading and inexpensive book

Great read and good overview.

This is short and mostly readable account of some of the main ideas in machine learning, including brief coverage of neural nets. It's well-organized and there are lots of useful examples and diagrams. Take the subtitle of this book ("an applied mathematics introduction") seriously. The author says in the prologue that the book is targeted to a narrow audience of applied mathematicians. The author jumps back and forth between very high-level discussion and hand-wavy technical discussion that assumes background in at least multivariable calculus and linear algebra. Does this work well for applied mathematicians? I think it will disappoint ML beginners without enough math background, and those with some ML experience who are looking for something concise and rigorous. One example is the brief PCA discussion, which assumes you will instantly see that what is needed is to get the eigenvectors of the covariance matrix. Another is the discussion of lasso regression, which suggests that the reader should make a plot and "draw comparisons between minimizing the loss function with penalty term and minimizing the loss function with a constant". (The book Intro to Statistical Learning with R actually explains this connection.) In some parts of the book I like the balance between intuition and technical detail. For example, I like the discussion of bias and variance (but have never found those target pictures very helpful). Some rigorous and concise books on machine learning that I like include Tom Mitchell's class Machine Learning and Hal Daumé III's A Course in Machine Learning.

The author gives a very good review of machine learning in theory or from an algorithmic point of view. You don't see a single line of code, but you will be very familiar with the concepts implemented in ML packages like Sci-kit learn. Actually, it'll help to understand what's done in Python. If Sci-kit learn package is a Python library, this book will help "to explain what the code is doing" (page 7). I think the people who knows ML well can learn a lot from this short book - it's relevant and up to date. The writing style is straightforward and fun to read!

When I started out, I ran several trading desks on the financial futures floors at the CME and CBOT. Fundamental and technical analysis were all that existed. I found that the only way to learn the quantitative aspect of the markets (circa 1983) was by walking around the exchange floors right after the close, picking up research/strategy papers off the floor near the most quantitatively-oriented firms. Fortunately for us, books authored by Dr. Wilmott and others like him have shed a light into the math, minds, and methodology of one of the most interesting areas of global markets.

Great book on intuition behind broad spectrum of Machine Learning approaches, full of practical examples. In fact, it is the only book aside from the Elements of Statistical Learning that I would recommend (and own). It is in strike contrast to the plethora of ML books on the market that are either too math heavy with little practical examples, or just show you how to apply python or R packages. Finally, entertaining value of this book should not be overlooked, not P. G. Wodehouse but close.

Todo el tema de Deep Learning lo zanja el autor en un párrafo, en la página 170. Más o menos en plan (y no cito literalmente por brevedad): "habréis escuchado frecuentemente la frase Deep Learning, nadie se pone de acuerdo en qué significa, y en lo que se ponen de acuerdo es que hay redes neurales y muchas capas ocultas". Entiendo que el autor va de tío enrollado, pero aquí se ha pasado tres pueblos.

The book I received is quite new and looks authentic. Though it was a bit damage from corner but I am satisfied with the product.seller looks honest as it sell original products.

For me, this book fills several gaps in my understanding of machine learning (ML) topics. It is my introduction to ML concepts, methods, definitions, jargon etc. What's more important, and why I give this book 5 stars, is that there is no programming code in it. This means that the book focuses not on learning programming tools (like many articles online) but on the actual subjects and the math which ML is based on. The examples come from a wide variety of areas and problems, which encourages the reader to think broader on how to apply ML techniques. Dr. Wilmott has a lot of humor which makes the book an enjoyable and interesting read. I consider this book a solid foundation, and most importantly an inspiration for me to start exploring ML properly. I most definitely recommend this book!

Too complicated for an ‘introduction’ book. It could have had more explanations of ideas to make them easier to understand.

Very nice introduction to machine learning, applied mathematicians will appreciate it. And one of the few books on machine learning (maybe the only one) to suggest the use of Matthews Correlation as a measure of performance of a classifier.