Great news. I'm particularly excited to see that they're implementing more thorough builtins for working with ANNs and other ML abstractions.
The current selection of layers is biased a bit towards vision, but my colleague and I are working on recurrent networks as we speak, which will unlock networks that operate on text, audio, and other data of a sequential or temporal character. Hopefully that will land in a few months with 11.1.
Awesome! Good to hear that there will be abstractions for general data processing too, and not just audio/video/text.
reference.wolfram.com seems to be down as I write this, but I'll check back in a few hours.
If you're working on Mathematica as your comment implies, thanks for your contributions. It's a wonderful piece of software (even if closed source)!