mechanoChemML.src.hparameters_dnn_grid

Module Contents

Classes

HyperParametersDNN This class is created to perform hyper-parameters search for DNNs.

Functions

getlist_str(option, sep=’,’, chars=None) Return a list from a ConfigParser option. By default,
getlist_int(option, sep=’,’, chars=None) Return a list from a ConfigParser option. By default,
mechanoChemML.src.hparameters_dnn_grid.getlist_str(option, sep=', ', chars=None)[source]

Return a list from a ConfigParser option. By default, split on a comma and strip whitespaces.

mechanoChemML.src.hparameters_dnn_grid.getlist_int(option, sep=', ', chars=None)[source]

Return a list from a ConfigParser option. By default, split on a comma and strip whitespaces.

class mechanoChemML.src.hparameters_dnn_grid.HyperParametersDNN(config, input_shape=1, output_shape=1, uniform_sample_number=10, neighbor_sample_number=5, iteration_time=3, sample_ratio=0.3, best_model_number=10, max_total_parameter=100000.0, repeat_train=3, debug=False)[source]

This class is created to perform hyper-parameters search for DNNs.

initialize_info(self)[source]
update_model_info(self, index0, history)[source]

update the summary for a particular DNN

compute_total_parameters_per_DNN(self, layers, neurons)[source]

Here, identical neurons per layer is assumed.

generate_all_DNNs(self)[source]
do_the_sampling(self)[source]
update_index_list(self)[source]
get_next_model(self)[source]
plot_best_models(self, model_number_to_plot=5)[source]
plot_all_models(self)[source]
mechanoChemML.src.hparameters_dnn_grid.config =[source]