What is PRML?
PRML is a book about mathematics in machine learning.
I read in turn with my friends using Translated one. Translated one is divided into two. We read Volume 1 in this summer vacation (2018).
Below is the topics of Volume 1
Chapter 1: Introduction
Chapter 2: Probability distribution
Chapter 3: Linear regression model
Chapter 4: Linear discriminant model
Chapter 5: Neural net work
Why I read this book with my friends
I read this for three reasons.
Firstly, I wanted to know the application example of the mathematics. I like math but I did not know how it is used.
Secondly, I wanted to know the inside of the artificial intelligence. AI is recently used in a lot of systems in the society. However, almost everyone do not know what AI can do and what AI cannot do. I thought that knowing inside is the best way to know it.
Thirdly, I wanted to improve my stamina for reading difficult books. I thought however difficult the book is, reading with friends would help me to complete it.
What I did
I made a presentation for chapter 2, which is about probability distribution.
What I learnt
As I wrote before, my goal is
- know the application example of mathematics
- know what AI can do and cannot do
- have stamina for reading a difficult book
First one and third one were accomplished. I studied a lot from my friends’ attitude for reading between lines. If I read it alone, I would give it up to read between lines at some places.
In terms of mathematics, we used Euler-Lagrange equation a lot. It was first time for me to do bayesian estimation, so I had some troubles to understand the flow of solving the optimization problems. In addition, I felt the lack of knowledge about the linear algebra. In the university classes. I learnt some operations of matrix but I did not know when it is used. It was great opportunity for learning it to read this book.
We are going to read volume 2 in spring vacation in 2019. I wish I could accomplish the second aim, “learning what AI can do and cannot do”.