- Add extract_models.jl with get_mlj_model_type and extract_model_from_machine methods.
- Modification of DenseNet neural networks sub-layers name.
- Type neuralode modification.
- Add name on neural networks sub-layers.
- Tests improvement.
- Add ComputationalResources on test neural_network_builder_test.jl.
- Tests improvement.
- Add rknn1, rknn2, rknn4 neural networks.
- Add DenseDiffEqFlux, DenseIdentityOut, DenseIdentityIn and DenseSampleTime struct.
- Add gaussian activation function for rbf.
- Improve mlj_model_interface for neuralODE training.
- Remove neuralode_type1 and keep only neuralode_type2 as neuralODE.
- Change folders structure.
- Tests improvement.
- Julia code formatter.
- Add Github Actions with continuous integration yml.
- Improve tests for continuous integration.
- Tests improvement.
- Removing GPU acceleration of neuralnetode_type1 and neuralnetode_type2.
- Add loss functions : MSE, RMSE, MAPE.
- Add loss functions tests.
- Improved args kws on black_box_identification.jl.
- Fixed Identification to AutomationLabsIdentification on architectures_neural_test.jl.
- Add [compat] for CUDA, DataFrames, DiffEqFlux, DifferentialEquations, Flux, FluxOptTools, MLJ, MLJFlux, MLJMultivariateStatsInterface, MLJParticleSwarmOptimization, MLUtils, MultivariateStats, Optim, ProgressMeter, StableRNGs, StatsBase.
- Fixed Identification to AutomationLabsIdentification.
- Initial public release.