I’m trying to do some logistic regression following a popular Machine Learning course from Andrew Ng.
In the code exercises, he uses the function
fminunc to find the minimum of the cost function, which seems to implement an BFGS algorithm for optimization.
I was looking for something similar in the Nx and Axon libraries, but haven’t found anything similar. I have seen the
Axon.Optimize module but don’t know if we can use the algorithms provided there as standalone optimization algorithms or are meant to be used only in the context of neural network training.