Reference

FluxNLPModel{T, S, C } <: AbstractNLPModel{T, S}

Data structure that makes the interfaces between neural networks defined with Flux.jl and NLPModels. A FluxNLPModel has fields

Arguments

• meta and counters retain informations about the FluxNLPModel;
• chain is the chained structure representing the neural network;
• data_train is the complete data training set;
• data_test is the complete data test;
• size_minibatch parametrizes the size of an training and test minibatches
• training_minibatch_iterator is an iterator over an training minibatches;
• test_minibatch_iterator is an iterator over the test minibatches;
• current_training_minibatch is the training minibatch used to evaluate the neural network;
• current_minibatch_test is the current test minibatch, it is not used in practice;
• w is the vector of weights/variables;

FluxNLPModel(chain_ANN data_train=MLDatasets.MNIST.traindata(Float32), data_test=MLDatasets.MNIST.testdata(Float32); size_minibatch=100)

Build a FluxNLPModel from the neural network represented by chain_ANN. chain_ANN is built using Flux.jl for more details. The other data required are: an iterator over the training dataset data_train, an iterator over the test dataset data_test and the size of the minibatch size_minibatch. Suppose (xtrn,ytrn) = Fluxnlp.data_train

accuracy(nlp::AbstractFluxNLPModel)

Compute the accuracy of the network nlp.chain on the entire test dataset. data_loader can be overwritten to include other data, device is set to cpu

minibatch_next_test!(nlp::AbstractFluxNLPModel)

Selects the next minibatch from nlp.test_minibatch_iterator. Returns the new current status of the iterator nlp.current_test_minibatch. minibatch_next_test! aims to be used in a loop or method call. if return false, it means that it reach the end of the mini-batch

minibatch_next_train!(nlp::AbstractFluxNLPModel)

Selects the next minibatch from nlp.training_minibatch_iterator. Returns the new current status of the iterator nlp.current_training_minibatch. minibatch_next_train! aims to be used in a loop or method call. if return false, it means that it reach the end of the mini-batch

reset_minibatch_test!(nlp::AbstractFluxNLPModel)

If a data_loader (an iterator object is passed to FluxNLPModel) then Select the first test minibatch for nlp.

reset_minibatch_train!(nlp::AbstractFluxNLPModel)

If a data_loader (an iterator object is passed to FluxNLPModel) then Select the first training minibatch for nlp.

set_vars!(model::AbstractFluxNLPModel{T,S}, new_w::AbstractVector{T}) where {T<:Number, S}

Sets the vaiables and rebuild the chain

update_type!(nlp::AbstractFluxNLPModel{T, S}, w::AbstractVector{V}) where {T, V, S}

Sets the variables and rebuild the chain to a specific type defined by weights.

g = grad!(nlp, w, g)

Evaluate ∇f(w), the gradient of the objective function at w in place.

Arguments

• nlp::AbstractFluxNLPModel{T, S}: the FluxNLPModel data struct;
• w::AbstractVector{V}: is the vector of weights/variables. The use of V allows for flexibility in specifying different precision types for weights and models.
• g::AbstractVector{}: the gradient vector.

Output

• g: the gradient at point w.

f = obj(nlp, w)

Evaluate the objective function f(w) of the non-linear programming (NLP) problem at the point w. 
If the precision of w and the precision expected by the nlp are different, ensure that the type of nlp.w matches the precision required by w.

Arguments

• nlp::AbstractFluxNLPModel{T, S}: the FluxNLPModel data struct;
• w::AbstractVector{V}: is the vector of weights/variables. The use of V allows for flexibility in specifying different precision types for weights and models.

Output

• f_w: the new objective function.

objgrad!(nlp, w, g)

Evaluate both f(w), the objective function of nlp at w, and ∇f(w), the gradient of the objective function at w in place.

Arguments

• nlp::AbstractFluxNLPModel{T, S}: the FluxNLPModel data struct;
• w::AbstractVector{V}: is the vector of weights/variables. The use of V allows for flexibility in specifying different precision types for weights and models.
• g::AbstractVector{V}: the gradient vector.

Output

• f_w, g: the new objective function, and the gradient at point w.