How to understand machine learning if you are not a great mathematician
A few months ago I studied projects in which incredible things are successfully realized thanks to machine learning.
And I caught fire with this. He said that I want to learn this. It doesn’t matter how hard it will be for me. I want to learn, and I will learn.
Let’s be honest: we all heard about the salaries of machine learning engineers. Take a look at this.
Impressive, right? But machine learning still needs to be mastered – and here the gloom begins.
Inspired, I began to study works on this topic, and you know what? Everywhere – math! Heaped up equations, linear algebra, vectors and strange characters.
That evening I cried like a child. But, as a good techie, he wiped away his tears and decided to study on his own.
Yes, I’m just another nerd trying to master machine learning.
But I’m bored of learning complex topics. Especially during quarantine. So I want to try something else. I will describe my learning process.
I will try.
Course of study
Mathematics → Statistics → Programming → Machine Learning → Amateur projects
When you search YouTube for videos about machine learning, you’ll definitely come across 3 main ones – from Siral Raval, Jabril and Daniel bourke.
All of them are beyond praise. Therefore, I decided to take the best from these videos.
Mathematics
There is a lot of debate about how well you need to know math to master machine learning. But you need to know for sure.
Perhaps some of you are damn ingenious in mathematics, and you only need to remember a few things. But most mere mortals like me need to learn everything from scratch.
Well, what exactly do you need to know? Just linear algebra and matanalysis.
I remind you: I am not a genius in mathematics. I am not good at math. I failed math analysis in all courses at the university!
So, is it possible to master the theory of machine learning without being a genius in mathematics?
Of course.
There is one caveat. If you are not friends with numbers, then this is because you do not understand the basics.
Remember the basics? About basics of linear algebra and mathematical analysis tells on the channel 3Blue1Brown Grant Sanderson. He needs to be given the Nobel Prize in education. He just takes the math, explains it in amazing form. Like a child. It’s fine.
So, my first step was to understand the basics of linear algebra and mathematical analysis. Believe me, after that, everything is much simpler.
We watched and comprehended these videos, now it’s time to put our knowledge into practice – on the course linear algebra from the largest specialist in the teaching of mathematics – Gilbert Strang of the Massachusetts Institute of Technology.
Just think: getting the same education as students who paid thousands of dollars for a full-time course! Yes, there will be no diploma from one of the best universities in the world, but accumulated knowledge is what ultimately matters.
Well, we learned this long course and practiced, now it’s the turn of mathematical analysis. The Khan Academy has amazing program, which gives everything you need in order to feel confident when dealing with sophisticated equations.
Statistics
Many people are confused by how much machine learning is similar to statistics. In fact, they are closely related to each other, so statistics are the key to understanding machine learning theory.
So focus and learn.
And to facilitate this task – a free course Probability – The Science of Uncertainty and Data from the Massachusetts Institute of Technology.
When reading the curriculum, you might think that the course is basic, but it is not. It covers enough to provide a basis for understanding probability theory. For everyone who likes to learn, here’s another course – Statistics and Probability from the Khan Academy. This is in addition, so relax.
Programming
If you, like me, are a software engineer, then for you now will be the most interesting.
The programming language you need to know is Python. The king of machine learning. Its simplicity makes the process of mastering the material very easy – at least at first.
I assume that you know programming, so I do not want to retell the contents of the courses for learning Python – there are many of them. In addition, there are excellent books. You decide where to gain knowledge.
It may be more convenient for someone to study the documentation or use a subscription to the online training platform, while someone else has a favorite teacher on Udemy. Above all, do not forget to practice in order to better understand what happens when programming for machine learning.
Well, let’s say you don’t know programming, and this will be your first line of code. In that case, I would choose Datacamp. Feel free to explore the topic yourself and see their Python course.
Machine learning
We are already far advanced. We studied mathematics, statistics, algorithms, cried for several nights. All for the sake of this moment.
Machine Learning Course from Andrew Eun – probably one of the best on the topic. It is not for beginners, so do not go far away your notes. Finally, how machine learning algorithms work will turn into a complete picture for you.
Another resource is Introduction to Machine Learning for Coders. A good course with detailed explanations of machine learning algorithms.
I advise you to go through both, study the issue from different angles, then you can tell which course turned out to be the most understandable.
I cannot but mention another program, which is very much praised. But it’s paid: it’s Introduction to Machine Learning Course on Udacity. If you have put aside some money and you are ready to invest in yourself, then this is a suitable case, but decide for yourself.
Amateur projects
Now you already know machine learning, but this is not enough. You need more practice. The book will help you here. Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow.
After that, you can take up amateur projects, but with the best machine learning libraries. If you, like me, do not like to rely on libraries without understanding what’s what, then don’t worry: you already understand. Therefore, I give this book at the very end.
And finally
- Before finishing, I want to give some tips.
- Do not be afraid to change the sequence of some steps. You might first want to learn Python, then linear algebra, and only then statistics.
- After learning something, practice more.
- Play with machine learning algorithms at any stage to figure out what’s what. Adjust settings to see what happens. Curiosity should be your weapon.
- Be patient. I know that all this takes time, it’s difficult. But it’s worth it.
- Kaggle to help you. Lots of Kaggle!
- Enjoy the process, not the result.
