Allow user to specify loss function using LossFunctions.jl

[dahong67]

It would be nice to also allow users to pass in losses from LossFunctions.jl: https://github.com/JuliaML/LossFunctions.jl

Currently, we define the following method signature:

GCPDecompositions.jl/src/gcp-opt.jl

Lines 17 to 18 in cfd310f

	gcp(X::Array, r, func=(x, m) -> (m - x)^2, grad=(x, m) -> ForwardDiff.derivative(m -> func(x, m), m), lower=-Inf) =
	_gcp(X, r, func, grad, lower, (;))

The simplest way to support LossFunctions.jl is probably to define an additional method signature as follows:

gcp(X::Array, r, loss::SupervisedLoss, lower=-Inf) = _gcp(X, r, ???, ???; lower)

where the ??? are what need to be filled in.

@alexmul1114, to start you may want to read through the LossFunctions.jl docs and try some examples. The following parts will be particularly relevant (using v0.10 docs since docs for current version appear to currently be missing the docstrings):

• https://juliaml.github.io/LossFunctions.jl/v0.10/user/interface/#Computing-the-Values
• https://juliaml.github.io/LossFunctions.jl/v0.10/user/interface/#Computing-the-1st-Derivatives

[dahong67]

Closed by #11