My Approach

I'll take a practical, top-down approach, starting with high-level frameworks with a focus on Deep Learning. I'll start with csv files, then work with images, text, sound, videos and then games. I'll work on increasingly difficult problems, beginning with test problems with clean datasets and the move towards real-world problems.

I started with Andrew Ng’s Course, but I lost motivation after three weeks. I didn’t have the desire to learn an new programming language which is not considered the standard, learn linear algebra, and combining it with machine learning notions.

My next stop was Kaggle campion Jeremy Howard’s fast AI course. A practical deep learning course starting with high level concepts. This has become a rough roadmap for me, but I was overwhelmed by the all the concepts they packed in each lesson. I grew an increased familiarity with the concepts, but I couldn’t apply it from scratch.

I realized tutorials are great at explaining concepts and for installing software, but they are not ideal to learn from. They are often created with the mindset of teaching you something. Instead of focusing on explanations this tutorial cuts out what is necessary for you to start learning. At heart, it’s just a few starting blocks and a curation of playgrounds to get your hands dirty.

This approach is inspired by this medium article and this quora post.