Also, just for reference, the MLJFlux package provides an interface between Flux and MLJ.
For what it's worth, I don't think we'll actually need to use the MLJFlux package. I think we first need to figure out how to construct Soss models that include Flux neural networks inside the Soss model. Once we do that, we should be pretty much done, and we can write up some tutorials for SossMLJ.
DilumAluthge
commented
4 years ago
I think that Bayesian neural networks would be a natural successor to our tutorials on Bayesian GLMs.
As far as the neural network library to use, I think Flux makes the most sense.
Prior art:
Also, just for reference, the MLJFlux package provides an interface between Flux and MLJ.
For what it's worth, I don't think we'll actually need to use the MLJFlux package. I think we first need to figure out how to construct Soss models that include Flux neural networks inside the Soss model. Once we do that, we should be pretty much done, and we can write up some tutorials for SossMLJ.