There are many great machine learning tutorials and courses out there. Some are really great at explaining complex themes. Some are inspiring to take action. Here are my favorite tutorials and a course that looks great on a resume.
When I started my journey in machine learning I hit a brick wall pretty fast. This wall was called MATH. All machine learning algorithms are in principle mathematical constructs.
Math is scary.
Sometimes I think most tutorials and even shool subjects start the wrong way around. Sure sometimes you need to know the basics first and work your way up. But up is the hard way. Learning (some of the) math for machine learning got much easier as soon as I started to develop ML models.
I had examples for the usage of the mathematic concepts and statistical methods. Not only did I know what a median was but also when it was used and why I needed it.
There are some tutorials that just work the other way around and inspire you to learn deeper from there. Here are my favorites and courses that help you.
machinelearningmastery.com
This blog got me into machine learning when I thought it was too hard to try. There are so many cool and easy examples that you can play around with and many articles to get you deeper down the rabbit hole. The author Jason Brownlee also sells books that provide a compact collection of tutorials and example code to get you started. Each one I bought was worth the money.
Alice in Wonderland text generation
This one got me going. it was so much fun to play around with. You’ll learn how to train a Long-Short-Term-Memory Network to continue a sentence from Alice in Wonderland. I later used the same basic algorithm to predict stock markets. It wasn’t much of a success, but I learned a lot.
MNIST Dataset – handwritten digit recognition
A great explanation for the classic image recognition example. It is still the go-to guide for me when I have an image recognition task to do. I always start here and go on further from there. This is where I first understood some basics of convolution and some quirks of Keras(like what dimensions to use on the input arrays.)
introduction-to-reinforcement-learning
Reinforcement Learning is one of the harder ML methods around. Before I found the tutorial I did a course on it and failed miserably. It was all about the functions and pagewise formulas without much context. I just can’t learn like that.
Then came this tutorial by Thomas Simonini. It promised to teach you to make your ML model play doom and sonic the hedgehog. That’s how it’s done!
There are many great tutorials on medium. But this one is just the best I found so far. After completing the first 3 parts of it I was able to write my own reinforcement learning scripts in python without having checked a line of example code. If they ever make a book out of it I’ll buy it in an instant.
Microsoft Professional Programm For Artificial Intelligence
Ok, this one is a mixed bag. The course is thorough and if you make it through you can have a certificate by Microsoft showing everyone that you are capable of implementing basic intelligent solutions. That’s the good part.
I learned a lot doing this course. The quality of information was great, everything was useful, but the quality of presentation was overall poor. I learned that people who are good in a thing are not necessarily good at teaching it.
There were some good courses, but also ones where experts just clicked furiously through PowerPoint pages full of formulas. One course was held by a non-native speaker and I couldn’t understand a word he said. Also, there were some tests that asked about things not covered by the course. In one instance I nearly bit my keyboard because they asked about a technical term that was never even mentioned in the videos or anywhere else on the Internet as far as I can tell.
It’s also quite expensive. I wouldn’t recommend it to people who want to learn but to people who have learned and want to prove it with a big name on the certificate. You’ll learn a lot but I would argue that someone without prior experience wouldn’t make it. It looks great on a resume though and after that, you won’t be scared to take on any educational challenge.