A Beginners Review of Three Popular Courses on AI and Machine Learning

Online Courses for AI

Online Courses for AI

Hi everyone,

It’s been a while since my last post. I’ve been focusing on Andrew Ng’s free AI course on Coursera at the moment. Now that I have had several AI courses under my belt, I thought I would give a quick review on the pros and cons of each in case you are having a hard time deciding which courses to take. I will be reviewing the AI courses from the Super Data Science Team on Udemy, the Deep Learning Nanodegree from Udacity hosted by Facebook and Andrew Ng’s Machine Learning course on Coursera hosted by Stanford. Here we go!

The most well rounded course goes to…

The Deep Learning Nanodegree from Udacity hosted by Facebook was the most well rounded course of the three. I participated in the 2018-2019 Facebook Scholarship Challenge which hosted 10,000 students worldwide. Although not exactly the same, it was a condensed 2 month subset of the Deep Learning Nanodegree. The content overall was rich with videos that explained the intuition behind AI concepts at a very easy and consumable pace. Some math behind the AI concepts were highlighted but not the focus of the course. Community participation was huge, making it very easy to get your questions answered in any time zone. Course admins often encourage group meetings, study sessions and meetups so if you’re naturally an introvert, that extra push could be useful. The framework of choice for the course is PyTorch with Python and from what I can see, it is growing very quickly in popularity among Data Scientists.

One pro tip that was mentioned by another student is if you’re using Udacity from India, they offer an “All You Can Eat” style for the same price as one Nanodegree in the US. If you’re savvy with using VPNs to make your traffic originate from another region, and have the time to consume lessons at a fast pace, you technically could get multiple Nanodegrees for the price of one.

https://www.udacity.com/course/deep-learning-nanodegree–nd101

The best course for learning the fundamentals..

Andrew Ng’s Machine Learning course on Coursera hosted by Stanford is the best course for learning the fundamentals of AI concepts like minimizing cost functions and gradient descent. It is very heavy on the math, leaning a lot on calculus and linear algebra. I’m only on week 2 but so far I’m enjoying the deep dive on the math behind the concepts. Some may have a hard time wrapping their heads around how/why gradient descent works, especially when there are already libraries in frameworks like PyTorch that can be called to do this automatically. If that describes you, and you have a good background in calculus and linear algebra, I would highly recommend this course. The programming language used in the course is Octave/Matlab which has a very similar syntax to Python. The only downside is that Octave/Matlab are good for prototyping AI models but they not very good for transitioning to a production environment. PyTorch used to have this problem, but they have since solved those issues, which is why it is becoming very popular. Although there are student forums available, there are no community managers directly interacting with students to help encourage participation.

https://www.coursera.org/learn/machine-learning

The best course for intermediate programmers…

For those that have a lot of programming experience already, and just want to dive deep into specific AI models, the courses from the Super Data Science Team on Udemy are the way to go. They don’t spend too much time on math and heavily focus on the intuition of AI concepts with an even heavier focus on code examples. They provide a swath of datasets, coding activities and coding solutions downloadable from their website. If you’re a seasoned programmer, sometimes it is easier to learn by looking at code examples, essentially reverse engineering functions and then modifying them to suit your needs. If you want to go deep into a particular AI model like Generative Adversarial Networks for image style transfer, then you can find the corresponding course offered by the Super Data Science Team. The only downside, or upside depending on your perspective, is that you have to create your own plan for what you want to learn by selecting different courses. Compared to the Nanodegree from Udacity which gave you a good breadth of AI concepts and models, you would be better served knowing what you wanted to learn with these courses.

https://www.udemy.com/courses/search/?src=ukw&q=Kirill+Eremenko+AI

That’s it! Thanks for reading. I hope that helped to shed some light on what these courses have to offer. If you have any questions or feedback, feel free to leave a comment below.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.