Resources
Deep learning/AI Courses
There are two very popular and well respected “intro to AI” type courses that I can recommend. Both are aimed at lay people and taught by excellent educators. Many people have worked their way through one or both of these courses to get into the field.
Fast.ai - This course is split into two parts. Each lasts about 6 months if you work through them in thorough detail. (You can watch the videos for both parts of the course for free.) Part 1 takes a very practical approach to help you get up to speed doing machine learning work for your field as quickly as possible. It covers some theory, but is more concerned with application first. Part 2 is more theoretical and works through the process of building the code from Part 1. If you want to know how to do the practice of deep learning to process data, but aren’t as concerned with being a deep learning researcher or developing new algorithms, then Part 1 (classes 1-7) of this course is for you.
Deeplearning.io - This is a series of five courses available on Coursera, and all together they are intended to take four to six months at 10 to 15 hours per week. As of Spring 2020, this costs $50 / month (independent of how long you spend on the courses) for people in the US. Andrew Ng from Stanford teaches all of these courses and he slowly builds up the theory of deep learning from scratch. This course assumes knowledge of linear algebra (vectors and matrices) and basic programming (in Python). If you have taken calculus, that will make some of the formulas more obvious, but calculus is not required for the course. If you want to become a deep learning researcher, definitely consider starting with this course.
Deep learning/AI Technologies to learn
Python
Most AI work today is done in Python. If you don’t know how to program yet, python is a great language to learn in. If you know how to program but don’t know Python yet, you can learn it quickly. The first step in learning about AI today is learning a little Python. You don’t need a ton, just enough to write a basic program.
Jupyter
Jupyter Notebook is a very simple way to create documents with live python code in them. This tool is used by most people in AI/DL/ML today.
PyTorch
PyTorch is an easy to use library for building and training deep learning models. OpenAI recently announced that they are consolidating thier work on PyTorch
- This video is an excellent introduction to PyTorch
- Alternately, you can clone and work through this repo. If you do, then start with the foreword.ipynb notebook.