E-Books →Calculus for Machine Learning Understanding the Language of Mathematics

When we try to understand machine learning algorithms, it is quite difficult to avoid calculus. This book is to help you refresh the calculus you learned or give you a quick start on just enough calculus to move forward.
When calculus is brought up, the first thing that comes to mind for many is difficult math. However, evaluating a calculus problem is just following some rules to manipulate it. However, the most important thing in studying calculus is to remember the physical nature it represents. At times, calculus can be abstract but is often not hard to visualize.
So why must we learn about calculus in studying machine learning? In many machine learning algorithms, we have a goal of what the machine should do and we expect it would behave in a certain way. Calculus is the tool for us to model the algorithm behavior. It allows us to see how the behavior of the algorithm would change if a parameter is changed. It also gives us insight on which direction we should fine-tune the algorithm or whether the algorithm is achieving the best it can do, even if it doesn't perfectly fit the training data.
As a practitioner, you need to know calculus is a tool for modeling. After reading this book, you should know why we cannot use accuracy measure as a loss function in training a neural network, which is because accuracy is not a differentiable function. You will also be able to explain why a neural network of larger size will be trained disproportionately slower, by counting the number of differentiations we need to compute in back propagation. Furthermore, if you understand calculus, you can convert an idea of the machine learning algorithm into code.
This is a book on the theoretical side of machine learning, but it does not aim to be comprehensive. The objective of this book is to provide you the background to understand the API documentation, or other people's work on machine learning. This book is to provide you an overview so you can go deeper with more advanced calculus books if you would like to.