54 void save(
const filesystem::path&)
const;
57 void load(
const filesystem::path&);
61 NeuronSelection* get_neurons_selection()
const {
return neurons_selection.get(); }
62 InputsSelection* get_inputs_selection()
const {
return inputs_selection.get(); }
63 void set_neurons_selection(
const string&);
64 void set_inputs_selection(
const string&);
66 TrainingStrategy* training_strategy =
nullptr;
68 unique_ptr<NeuronSelection> neurons_selection;
70 unique_ptr<InputsSelection> inputs_selection;
void set(TrainingStrategy *new_training_strategy)
Definition model_selection.h:31
void load(const filesystem::path &)
Loads the model selection configuration from disk.
void save(const filesystem::path &) const
Saves the model selection configuration to disk.
NeuronsSelectionResults perform_neurons_selection()
Runs the configured neurons selection algorithm.
void from_JSON(const JsonDocument &)
Loads model selection configuration from a JSON document.
bool has_training_strategy() const
Definition model_selection.h:30
void check() const
Checks that the training strategy and its dependencies are consistent before running selection.
void set_default()
Restores default algorithms and parameters for inputs and neurons selection.
ModelSelection(TrainingStrategy *=nullptr)
Constructs a model selection bound to an optional training strategy.
const TrainingStrategy * get_training_strategy() const
Definition model_selection.h:29
InputsSelectionResults perform_input_selection()
Runs the configured inputs selection algorithm.
void to_JSON(JsonWriter &) const
Writes the current model selection configuration to a JSON writer.
Abstract base class for algorithms that select the optimal number of hidden neurons.
Definition neuron_selection.h:20
High-level orchestrator pairing a Loss with an Optimizer for a network/dataset.
Definition training_strategy.h:24
Definition adaptive_moment_estimation.h:14
Aggregated results of a neurons selection run including the optimal neuron count and error histories.
Definition neuron_selection.h:111