Earlier this month, I've started learning from a couple of Deep Learning tutorials/courses/books. I had been thinking of picking up Deep Learning again for some time, but could not push it into my schedule. I do have a background in Machine Learning and Data Science, but I have never used Deep Learning methods in production applications yet. I have studied them in the past. Now that the space has moved quite fast, and Deep Learning now dominates over rest of the methods in ML, I feel the need to bring myself up to speed with the whole field.

Last year, I did work on some experimental applications in drawing and sketching using tensorflow pose detection models. However I got distracted by other stuff, and never made a complete application.

This time I've started with a more structured approach. I've started with one course and one book - using some very different approaches:

Practical Deep Learning for Coders (fastai)

This is a course for programmers by Jeremy Howard of fastai, who has also co-authored a book on the topic. The course follows some of the material of the book. I have finished Chapter 1 of the course on YouTube and also run the notebooks associated, and have found the course very helpful. Although, I'm familiar with most of the material, but it is a good revision

The Little Learner

I pre-ordered this book early last year, and got it delivered to my Kindle. However I haven't had time to study it since. This book uses a dialectic approach of the Little series of books to teach Deep Learning. It uses Racket programming language - which I'm quite familiar with. One of the first open source applications I wrote was using PLT Scheme (which is the old name of Racket). The book is quite fun to follow but I'm only through Chapter 0 and 1 till now.

I'll use my blog to keep track of some of the key landmarks as I study this subject. I hope to make something with the newly studied material and I will write about it as I build it.