I came across a great episode while listening to Software Engineering Radio the other day while on my morning walk.
Episode 391 features Jeremy Howard from fast.ai explaining deep learning from concept to implementation.
I had heard of fast.ai before in an Elixir meetup where I met a gentlemen who designs machine learning systems for monitoring people in hospital beds. The system uses visual detection aided by deep learning to alert staff if a patient has fallen from their bed hopefully getting the patient the help they need more quickly. In our discussion he brought up fast.ai as a great learning resource.
The course is available at https://course.fast.ai/ and includes videos, jupyter notebooks, and instructions on how to use cloud resources to get up and running.
Here is a preview of lesson one if you want to get a quick peek: