MOOCs and Lecture Series

These are my picks for favorite Machine Learning MOOCs and lecture videos. While they aren’t updated often (or at all), I find myself coming back to these whenever I need to recap important concepts. If you’re looking for more frequently updated content, look here.

Source: Wikimedia Commons

• Machine Learning by Andrew Ng- The course that started it all, I won’t say any more.
• CS231 Convolutional Neural Networks for Visual Recognition- The legendary course by Fei-Fei Li and her students (used to be Andrej Karpathy). Excellent lecture notes that you can always come back to.
• Practical Deep Learning for Coders by fast.ai- A superb way of getting into modern Deep Learning implementation and research, this course espouses hands on learning over reading tomes of mathematics. In my opinion, this is the right way to learn a subject as complex and multifaceted as it’s very easy to get stuck in dependency hell, reading more and more until you lose sight of why you were reading something in the first place. Guilty as charged .
• Oxford Deep Learning for NLP- Cannot recommend this enough. Legends Phil Blumson and Chris Dyer build a solid foundation of concepts before gradually building up to modern deep learning architectures for NLP tasks such as Language Modeling, Machine Comprehension, Translation and Question Answering. It helps to practice coding small examples in parallel with watching the lecture videos.
• Reinforcement Learning by David Silver- Highly recommended introduction to RL. Follow along with the canonical textbook by Sutton and Barto and I promise you that you will be ready to understand modern research papers in the field. Silver is a phenomenal teacher, this is one of the best courses I have ever taken, period.
• Deep RL by Sergey Levine- This is not meant as an introductory course. You have been warned. That said, this course do provide a detailed treatment of optimal control and model based learning, which are really useful in robotics or really any application where you care about sample efficiency.

If you want to catch up on the latest happenings in the world of Deep Learning or Reinforcement Learning research, summer schools and bootcamps are great. One hour is often enough to get an overview of a topic, and fill up your reading list. The links below are a few years old, but still relevant as of 2018. I would however suggest finding more recent talks or newer versions of these summer schools.

• Deep RL Bootcamp 2017
• Deep Learning School 2016