Sounds cool! I wonder if this will be with TensorFlow. If so, does the RPis videocore have any capability to accelerate such things or not out of interest.
There's another one that goes into more detail on the "how" of running other algorithms: https://rpiplayground.wordpress.com/tag/raspberry-pi-gpu/
I'm not sure if there are limitations that would keep it from being interesting for TF or not; I don't know enough about it.