Companies give you learning days, which is awesome. This year I’m learning deep learning, because I think it just might become a big thing :D.
The fast AI course itself recommends you to blog about your experience on the course, so soon I will be deploying my first model (an identifier that can tell Zoro the Havenese apart from… other Havenese dogs, pretty cool - but I might need to take some photos of me holding other havenese dogs for the validation set!)
This course is fantastic, Jeremy explains things so clearly, I highly recommend this course if you want to build and understand these systems. Whilst I know a lot of the basics I’m enjoying going back through them and then picking up new things, like just how easy and fast it is to locally tune an image classifier. It’s hard to intuit how few examples you need to fine tune a well built image model until you try it.
It’s quite astonishing how far ANNs have come in the last 10-15 years. In 2014 I was sent to canary wharf to learn about them from a company (Matrica), that was selling us forecasting software using them. I’m pretty sure we were training on CPUs :o ! When I went back to my uni for a talk and asked students who knew what an ANN or ML was, almost no one in the class did. I imagine that numbers quite different now!