How to teach machine learning in a simple yet useful way? Which machine learning ideas can be presented with pen & paper? Which properties of machine learning are in fact properties of any learning, including by us, humans? How to approach various audiences - from programmers, through physicists to people without strong STEM background?
Machine learning (and especially deep learning) is being presented as something hard, requiring years of mathematics and computer science. I don't think it's the case. Many ideas are simple, extension from some common-sense intuitions and can be taught to people from various backgrounds.
I have already shared some things in my blog posts:
Well, it's not my first adventure with trying to teach complex ideas in a simple way, see Quantum mechanics for high-school students and Quantum Game with Photons. Later allows you to play with an actual simulation of a single photon, without need of learning quantum mechanics beforehand.