metaperceptron.core package

metaperceptron.core.base_mlp_numpy module

class metaperceptron.core.base_mlp_numpy.BaseMhaMlp(hidden_size=50, act1_name='tanh', act2_name='sigmoid', obj_name=None, optimizer='OriginalWOA', optimizer_paras=None, verbose=True)

Bases: BaseEstimator

Defines the most general class for Metaheuristic-based MLP model that inherits the BaseMlpNumpy class

Parameters:

• hidden_size (int, default=50) – The number of hidden nodes

• act1_name (str, default='tanh') – Activation function for the hidden layer. The supported activation functions are: [“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”, “silu”]

• act2_name (str, default='sigmoid') – Activation function for the hidden layer. The supported activation functions are: [“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”, “silu”]

• obj_name (None or str, default=None) – The name of objective for the problem, also depend on the problem is classification and regression.

• optimizer (str or instance of Optimizer class (from Mealpy library), default = "OriginalWOA") – The Metaheuristic Algorithm that use to solve the feature selection problem. Current supported list, please check it here: https://github.com/thieu1995/mealpy. If a custom optimizer is passed, make sure it is an instance of Optimizer class.

• optimizer_paras (None or dict of parameter, default=None) – The parameter for the optimizer object. If None, the default parameters of optimizer is used (defined in https://github.com/thieu1995/mealpy.) If dict is passed, make sure it has at least epoch and pop_size parameters.

• verbose (bool, default=True) – Whether to print progress messages to stdout.

CLS_OBJ_LOSSES = None

SUPPORTED_ACTIVATIONS = ['none', 'relu', 'leaky_relu', 'celu', 'prelu', 'gelu', 'elu', 'selu', 'rrelu', 'tanh', 'hard_tanh', 'sigmoid', 'hard_sigmoid', 'log_sigmoid', 'swish', 'hard_swish', 'soft_plus', 'mish', 'soft_sign', 'tanh_shrink', 'soft_shrink', 'hard_shrink', 'softmin', 'softmax', 'log_softmax', 'silu']

SUPPORTED_CLS_METRICS = {'AS': 'max', 'BSL': 'min', 'CEL': 'min', 'CKS': 'max', 'F1S': 'max', 'F2S': 'max', 'FBS': 'max', 'GINI': 'min', 'GMS': 'max', 'HL': 'min', 'HS': 'max', 'JSI': 'max', 'KLDL': 'min', 'LS': 'max', 'MCC': 'max', 'NPV': 'max', 'PS': 'max', 'ROC-AUC': 'max', 'RS': 'max', 'SS': 'max'}

SUPPORTED_CLS_OBJECTIVES = {'AS': 'max', 'BSL': 'min', 'CEL': 'min', 'CKS': 'max', 'F1S': 'max', 'F2S': 'max', 'FBS': 'max', 'GINI': 'min', 'GMS': 'max', 'HL': 'min', 'HS': 'max', 'JSI': 'max', 'KLDL': 'min', 'LS': 'max', 'MCC': 'max', 'NPV': 'max', 'PS': 'max', 'ROC-AUC': 'max', 'RS': 'max', 'SS': 'max'}

SUPPORTED_OPTIMIZERS = ['OriginalABC', 'OriginalACOR', 'AugmentedAEO', 'EnhancedAEO', 'ImprovedAEO', 'ModifiedAEO', 'OriginalAEO', 'MGTO', 'OriginalAGTO', 'DevALO', 'OriginalALO', 'OriginalAO', 'OriginalAOA', 'IARO', 'LARO', 'OriginalARO', 'OriginalASO', 'OriginalAVOA', 'OriginalArchOA', 'AdaptiveBA', 'DevBA', 'OriginalBA', 'DevBBO', 'OriginalBBO', 'OriginalBBOA', 'OriginalBES', 'ABFO', 'OriginalBFO', 'OriginalBMO', 'DevBRO', 'OriginalBRO', 'OriginalBSA', 'ImprovedBSO', 'OriginalBSO', 'CleverBookBeesA', 'OriginalBeesA', 'ProbBeesA', 'OriginalCA', 'OriginalCDO', 'OriginalCEM', 'OriginalCGO', 'DevCHIO', 'OriginalCHIO', 'OriginalCOA', 'OCRO', 'OriginalCRO', 'OriginalCSA', 'OriginalCSO', 'OriginalCircleSA', 'OriginalCoatiOA', 'JADE', 'OriginalDE', 'SADE', 'SAP_DE', 'DevDMOA', 'OriginalDMOA', 'OriginalDO', 'DevEFO', 'OriginalEFO', 'OriginalEHO', 'AdaptiveEO', 'ModifiedEO', 'OriginalEO', 'OriginalEOA', 'LevyEP', 'OriginalEP', 'CMA_ES', 'LevyES', 'OriginalES', 'Simple_CMA_ES', 'OriginalESOA', 'OriginalEVO', 'OriginalFA', 'DevFBIO', 'OriginalFBIO', 'OriginalFFA', 'OriginalFFO', 'OriginalFLA', 'DevFOA', 'OriginalFOA', 'WhaleFOA', 'OriginalFOX', 'OriginalFPA', 'BaseGA', 'EliteMultiGA', 'EliteSingleGA', 'MultiGA', 'SingleGA', 'OriginalGBO', 'DevGCO', 'OriginalGCO', 'OriginalGJO', 'OriginalGOA', 'DevGSKA', 'OriginalGSKA', 'Matlab101GTO', 'Matlab102GTO', 'OriginalGTO', 'GWO_WOA', 'IGWO', 'OriginalGWO', 'RW_GWO', 'OriginalHBA', 'OriginalHBO', 'OriginalHC', 'SwarmHC', 'OriginalHCO', 'OriginalHGS', 'OriginalHGSO', 'OriginalHHO', 'DevHS', 'OriginalHS', 'OriginalICA', 'OriginalINFO', 'OriginalIWO', 'DevJA', 'LevyJA', 'OriginalJA', 'DevLCO', 'ImprovedLCO', 'OriginalLCO', 'OriginalMA', 'OriginalMFO', 'OriginalMGO', 'OriginalMPA', 'OriginalMRFO', 'WMQIMRFO', 'OriginalMSA', 'DevMVO', 'OriginalMVO', 'OriginalNGO', 'ImprovedNMRA', 'OriginalNMRA', 'OriginalNRO', 'OriginalOOA', 'OriginalPFA', 'OriginalPOA', 'AIW_PSO', 'CL_PSO', 'C_PSO', 'HPSO_TVAC', 'LDW_PSO', 'OriginalPSO', 'P_PSO', 'OriginalPSS', 'DevQSA', 'ImprovedQSA', 'LevyQSA', 'OppoQSA', 'OriginalQSA', 'OriginalRIME', 'OriginalRUN', 'GaussianSA', 'OriginalSA', 'SwarmSA', 'DevSARO', 'OriginalSARO', 'DevSBO', 'OriginalSBO', 'DevSCA', 'OriginalSCA', 'QleSCA', 'OriginalSCSO', 'ImprovedSFO', 'OriginalSFO', 'L_SHADE', 'OriginalSHADE', 'OriginalSHIO', 'OriginalSHO', 'ImprovedSLO', 'ModifiedSLO', 'OriginalSLO', 'DevSMA', 'OriginalSMA', 'DevSOA', 'OriginalSOA', 'OriginalSOS', 'DevSPBO', 'OriginalSPBO', 'OriginalSRSR', 'DevSSA', 'OriginalSSA', 'OriginalSSDO', 'OriginalSSO', 'OriginalSSpiderA', 'OriginalSSpiderO', 'OriginalSTO', 'OriginalSeaHO', 'OriginalServalOA', 'OriginalTDO', 'DevTLO', 'ImprovedTLO', 'OriginalTLO', 'OriginalTOA', 'DevTPO', 'OriginalTS', 'OriginalTSA', 'OriginalTSO', 'EnhancedTWO', 'LevyTWO', 'OppoTWO', 'OriginalTWO', 'DevVCS', 'OriginalVCS', 'OriginalWCA', 'OriginalWDO', 'OriginalWHO', 'HI_WOA', 'OriginalWOA', 'OriginalWaOA', 'OriginalWarSO', 'OriginalZOA']

SUPPORTED_REG_METRICS = {'A10': 'max', 'A20': 'max', 'A30': 'max', 'ACOD': 'max', 'APCC': 'max', 'AR': 'max', 'AR2': 'max', 'CI': 'max', 'COD': 'max', 'COR': 'max', 'COV': 'max', 'CRM': 'min', 'DRV': 'min', 'EC': 'max', 'EVS': 'max', 'GINI': 'min', 'GINI_WIKI': 'min', 'JSD': 'min', 'KGE': 'max', 'MAAPE': 'min', 'MAE': 'min', 'MAPE': 'min', 'MASE': 'min', 'ME': 'min', 'MRB': 'min', 'MRE': 'min', 'MSE': 'min', 'MSLE': 'min', 'MedAE': 'min', 'NNSE': 'max', 'NRMSE': 'min', 'NSE': 'max', 'OI': 'max', 'PCC': 'max', 'PCD': 'max', 'R': 'max', 'R2': 'max', 'R2S': 'max', 'RAE': 'min', 'RMSE': 'min', 'RSE': 'min', 'RSQ': 'max', 'SMAPE': 'min', 'VAF': 'max', 'WI': 'max'}

SUPPORTED_REG_OBJECTIVES = {'A10': 'max', 'A20': 'max', 'A30': 'max', 'ACOD': 'max', 'APCC': 'max', 'AR': 'max', 'AR2': 'max', 'CI': 'max', 'COD': 'max', 'COR': 'max', 'COV': 'max', 'CRM': 'min', 'DRV': 'min', 'EC': 'max', 'EVS': 'max', 'GINI': 'min', 'GINI_WIKI': 'min', 'JSD': 'min', 'KGE': 'max', 'MAAPE': 'min', 'MAE': 'min', 'MAPE': 'min', 'MASE': 'min', 'ME': 'min', 'MRB': 'min', 'MRE': 'min', 'MSE': 'min', 'MSLE': 'min', 'MedAE': 'min', 'NNSE': 'max', 'NRMSE': 'min', 'NSE': 'max', 'OI': 'max', 'PCC': 'max', 'PCD': 'max', 'R': 'max', 'R2': 'max', 'R2S': 'max', 'RAE': 'min', 'RMSE': 'min', 'RSE': 'min', 'RSQ': 'max', 'SMAPE': 'min', 'VAF': 'max', 'WI': 'max'}

create_network(X, y)

evaluate(y_true, y_pred, list_metrics=None)

Return the list of performance metrics of the prediction.

Parameters:

• y_true (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• y_pred (array-like of shape (n_samples,) or (n_samples, n_outputs)) – Predicted values for X.

• list_metrics (list) – You can get metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

fit(X, y, lb=(-1.0,), ub=(1.0,), save_population=False, obj_weights=None)

static load_model(load_path='history', filename='model.pkl')

objective_function(solution=None)

predict(X, return_prob=False)

Inherit the predict function from BaseMlpNumpy class, with 1 more parameter return_prob.

Parameters:

• X ({array-like, sparse matrix} of shape (n_samples, n_features)) – The input data.

• return_prob (bool, default=False) –

  It is used for classification problem:

  • If True, the returned results are the probability for each sample

  • If False, the returned results are the predicted labels

save_evaluation_metrics(y_true, y_pred, list_metrics=('RMSE', 'MAE'), save_path='history', filename='metrics.csv')

Save evaluation metrics to csv file

Parameters:

• y_true (ground truth data) –

• y_pred (predicted output) –

• list_metrics (list of evaluation metrics) –

• save_path (saved path (relative path, consider from current executed script path)) –

• filename (name of the file, needs to have ".csv" extension) –

save_model(save_path='history', filename='model.pkl')

Save model to pickle file

Parameters:

• save_path (saved path (relative path, consider from current executed script path)) –

• filename (name of the file, needs to have ".pkl" extension) –

save_training_loss(save_path='history', filename='loss.csv')

Save the loss (convergence) during the training process to csv file.

Parameters:

• save_path (saved path (relative path, consider from current executed script path)) –

• filename (name of the file, needs to have ".csv" extension) –

save_y_predicted(X, y_true, save_path='history', filename='y_predicted.csv')

Save the predicted results to csv file

Parameters:

• X (The features data, nd.ndarray) –

• y_true (The ground truth data) –

• save_path (saved path (relative path, consider from current executed script path)) –

• filename (name of the file, needs to have ".csv" extension) –

score(X, y, method=None)

Return the metric of the prediction.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• method (str, default="RMSE") – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

result – The result of selected metric

Return type:

float

scores(X, y, list_methods=None)

Return the list of metrics of the prediction.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• list_methods (list, default=("MSE", "MAE")) – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

set_fit_request(*, lb: bool | None | str = '$UNCHANGED$', obj_weights: bool | None | str = '$UNCHANGED$', save_population: bool | None | str = '$UNCHANGED$', ub: bool | None | str = '$UNCHANGED$') → BaseMhaMlp

Request metadata passed to the fit method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to fit.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

• lb (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for lb parameter in fit.

• obj_weights (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for obj_weights parameter in fit.

• save_population (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for save_population parameter in fit.

• ub (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for ub parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, return_prob: bool | None | str = '$UNCHANGED$') → BaseMhaMlp

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to predict.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

return_prob (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for return_prob parameter in predict.

Returns:

self – The updated object.

Return type:

object

set_score_request(*, method: bool | None | str = '$UNCHANGED$') → BaseMhaMlp

Request metadata passed to the score method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to score.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

method (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for method parameter in score.

Returns:

self – The updated object.

Return type:

object

class metaperceptron.core.base_mlp_numpy.MlpNumpy(input_size=5, hidden_size=10, output_size=1, act1_name='tanh', act2_name='sigmoid')

Bases: object

This class defines the general Multi-Layer Perceptron (MLP) model using Numpy

Parameters:

• input_size (int, default=5) – The number of input nodes

• hidden_size (int, default=10) – The number of hidden nodes

• output_size (int, default=1) – The number of output nodes

• act1_name (str, default='tanh') – Activation function for the hidden layer. The supported activation functions are: [“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”, “silu”]

• act2_name (str, default='sigmoid') – Activation function for the hidden layer. The supported activation functions are: [“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”, “silu”]

fit(X, y)

Fit the model to data matrix X and target(s) y.

Parameters:

• X (ndarray or sparse matrix of shape (n_samples, n_features)) – The input data.

• y (ndarray of shape (n_samples,) or (n_samples, n_outputs)) – The target values (class labels in classification, real numbers in regression).

Returns:

self – Returns a trained MLP model.

Return type:

object

forward(inputs)

get_weights()

get_weights_size()

predict(X)

Predict using the Extreme Learning Machine model.

Parameters:

X ({array-like, sparse matrix} of shape (n_samples, n_features)) – The input data.

Returns:

y – The predicted values.

Return type:

ndarray of shape (n_samples, n_outputs)

set_weights(weights)

update_weights_from_solution(solution)

metaperceptron.core.base_mlp_torch module

class metaperceptron.core.base_mlp_torch.BaseMlpTorch(hidden_size=50, act1_name='tanh', act2_name='sigmoid', obj_name=None, max_epochs=1000, batch_size=32, optimizer='SGD', optimizer_paras=None, verbose=False)

Bases: BaseEstimator

Defines the most general class for traditional MLP models that inherits the BaseEstimator class of Scikit-Learn library.

Parameters:

• hidden_size (int, default=50) – The hidden size of MLP network (This network only has single hidden layer).

• act1_name (str, defeault="tanh") – This is activation for hidden layer. The supported activation are: {“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “silu”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”}.

• act2_name (str, defeault="sigmoid") – This is activation for output layer. The supported activation are: {“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “silu”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”}.

• obj_name (str, default=None) – The name of objective for the problem, also depend on the problem is classification and regression.

• max_epochs (int, default=1000) – Maximum number of epochs / iterations / generations

• batch_size (int, default=32) – The batch size

• optimizer (str, default = "SGD") – The gradient-based optimizer from Pytorch. List of supported optimizer is: [“Adadelta”, “Adagrad”, “Adam”, “Adamax”, “AdamW”, “ASGD”, “LBFGS”, “NAdam”, “RAdam”, “RMSprop”, “Rprop”, “SGD”]

• optimizer_paras (dict or None, default=None) – The dictionary parameters of the selected optimizer.

• verbose (bool, default=True) – Whether to print progress messages to stdout.

CLS_OBJ_LOSSES = None

SUPPORTED_CLS_METRICS = {'AS': 'max', 'BSL': 'min', 'CEL': 'min', 'CKS': 'max', 'F1S': 'max', 'F2S': 'max', 'FBS': 'max', 'GINI': 'min', 'GMS': 'max', 'HL': 'min', 'HS': 'max', 'JSI': 'max', 'KLDL': 'min', 'LS': 'max', 'MCC': 'max', 'NPV': 'max', 'PS': 'max', 'ROC-AUC': 'max', 'RS': 'max', 'SS': 'max'}

SUPPORTED_LOSSES = {'MAE': <class 'torch.nn.modules.loss.L1Loss'>, 'MSE': <class 'torch.nn.modules.loss.MSELoss'>}

SUPPORTED_OPTIMIZERS = ['Adadelta', 'Adagrad', 'Adam', 'Adamax', 'AdamW', 'ASGD', 'LBFGS', 'NAdam', 'RAdam', 'RMSprop', 'Rprop', 'SGD']

SUPPORTED_REG_METRICS = {'A10': 'max', 'A20': 'max', 'A30': 'max', 'ACOD': 'max', 'APCC': 'max', 'AR': 'max', 'AR2': 'max', 'CI': 'max', 'COD': 'max', 'COR': 'max', 'COV': 'max', 'CRM': 'min', 'DRV': 'min', 'EC': 'max', 'EVS': 'max', 'GINI': 'min', 'GINI_WIKI': 'min', 'JSD': 'min', 'KGE': 'max', 'MAAPE': 'min', 'MAE': 'min', 'MAPE': 'min', 'MASE': 'min', 'ME': 'min', 'MRB': 'min', 'MRE': 'min', 'MSE': 'min', 'MSLE': 'min', 'MedAE': 'min', 'NNSE': 'max', 'NRMSE': 'min', 'NSE': 'max', 'OI': 'max', 'PCC': 'max', 'PCD': 'max', 'R': 'max', 'R2': 'max', 'R2S': 'max', 'RAE': 'min', 'RMSE': 'min', 'RSE': 'min', 'RSQ': 'max', 'SMAPE': 'min', 'VAF': 'max', 'WI': 'max'}

create_network(X, y)

evaluate(y_true, y_pred, list_metrics=None)

Return the list of performance metrics of the prediction.

Parameters:

• y_true (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• y_pred (array-like of shape (n_samples,) or (n_samples, n_outputs)) – Predicted values for X.

• list_metrics (list) – You can get metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

fit(X, y)

static load_model(load_path='history', filename='model.pkl')

predict(X, return_prob=False)

Inherit the predict function from BaseMlp class, with 1 more parameter return_prob.

Parameters:

• X ({array-like, sparse matrix} of shape (n_samples, n_features)) – The input data.

• return_prob (bool, default=False) –

  It is used for classification problem:

  • If True, the returned results are the probability for each sample

  • If False, the returned results are the predicted labels

save_evaluation_metrics(y_true, y_pred, list_metrics=('RMSE', 'MAE'), save_path='history', filename='metrics.csv')

Save evaluation metrics to csv file

Parameters:

• y_true (ground truth data) –

• y_pred (predicted output) –

• list_metrics (list of evaluation metrics) –

• save_path (saved path (relative path, consider from current executed script path)) –

• filename (name of the file, needs to have ".csv" extension) –

save_model(save_path='history', filename='model.pkl')

Save model to pickle file

Parameters:

• save_path (saved path (relative path, consider from current executed script path)) –

• filename (name of the file, needs to have ".pkl" extension) –

save_training_loss(save_path='history', filename='loss.csv')

Save the loss (convergence) during the training process to csv file.

Parameters:

• save_path (saved path (relative path, consider from current executed script path)) –

• filename (name of the file, needs to have ".csv" extension) –

save_y_predicted(X, y_true, save_path='history', filename='y_predicted.csv')

Save the predicted results to csv file

Parameters:

• X (The features data, nd.ndarray) –

• y_true (The ground truth data) –

• save_path (saved path (relative path, consider from current executed script path)) –

• filename (name of the file, needs to have ".csv" extension) –

score(X, y, method=None)

Return the metric of the prediction.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• method (str, default="RMSE") – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

result – The result of selected metric

Return type:

float

scores(X, y, list_methods=None)

Return the list of metrics of the prediction.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• list_methods (list, default=("MSE", "MAE")) – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

set_predict_request(*, return_prob: bool | None | str = '$UNCHANGED$') → BaseMlpTorch

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to predict.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

return_prob (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for return_prob parameter in predict.

Returns:

self – The updated object.

Return type:

object

set_score_request(*, method: bool | None | str = '$UNCHANGED$') → BaseMlpTorch

Request metadata passed to the score method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to score.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

method (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for method parameter in score.

Returns:

self – The updated object.

Return type:

object

class metaperceptron.core.base_mlp_torch.MlpTorch(input_size, hidden_size, output_size, act1_name='tanh', act2_name='sigmoid')

Bases: Module

Define the MLP model

SUPPORTED_ACTIVATIONS = ['none', 'threshold', 'relu', 'rrelu', 'hardtanh', 'relu6', 'sigmoid', 'hardsigmoid', 'tanh', 'silu', 'mish', 'hardswish', 'elu', 'celu', 'selu', 'glu', 'gelu', 'hardshrink', 'leakyrelu', 'logsigmoid', 'softplus', 'softshrink', 'multiheadattention', 'prelu', 'softsign', 'tanhshrink', 'softmin', 'softmax', 'logsoftmax']

forward(x)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

training: bool

metaperceptron.core.mha_mlp module

class metaperceptron.core.mha_mlp.MhaMlpClassifier(hidden_size=50, act1_name='tanh', act2_name='sigmoid', obj_name='CEL', optimizer='OriginalWOA', optimizer_paras=None, verbose=True)

Bases: BaseMhaMlp, ClassifierMixin

Defines the general class of Metaheuristic-based MLP model for Classification problems that inherit the BaseMhaMlp and ClassifierMixin classes.

Parameters:

• hidden_size (int, default=50) – The number of hidden nodes

• act1_name (str, default='tanh') – Activation function for the hidden layer. The supported activation functions are: [“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”, “silu”]

• act2_name (str, default='sigmoid') – Activation function for the hidden layer. The supported activation functions are: [“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”, “silu”]

• obj_name (str, default="MSE") – Current supported objective functions, please check it here: https://github.com/thieu1995/permetrics

• optimizer (str or instance of Optimizer class (from Mealpy library), default = "OriginalWOA") – The Metaheuristic Algorithm that use to solve the feature selection problem. Current supported list, please check it here: https://github.com/thieu1995/mealpy. If a custom optimizer is passed, make sure it is an instance of Optimizer class.

• optimizer_paras (None or dict of parameter, default=None) – The parameter for the optimizer object. If None, the default parameters of optimizer is used (defined in https://github.com/thieu1995/mealpy.) If dict is passed, make sure it has at least epoch and pop_size parameters.

• verbose (bool, default=True) – Whether to print progress messages to stdout.

Examples

>>> from metaperceptron import Data, MhaMlpClassifier
>>> from sklearn.datasets import make_classification
>>> X, y = make_classification(n_samples=100, random_state=1)
>>> data = Data(X, y)
>>> data.split_train_test(test_size=0.2, random_state=1)
>>> data.X_train_scaled, scaler = data.scale(data.X_train, method="MinMaxScaler")
>>> data.X_test_scaled = scaler.transform(data.X_test)
>>> opt_paras = {"name": "GA", "epoch": 10, "pop_size": 30}
>>> print(MhaMlpClassifier.SUPPORTED_CLS_OBJECTIVES)
{'PS': 'max', 'NPV': 'max', 'RS': 'max', ...., 'KLDL': 'min', 'BSL': 'min'}
>>> model = MhaMlpClassifier(hidden_size=50, act1_name="tanh", act2_name="sigmoid",
>>>             obj_name="NPV", optimizer="OriginalWOA", optimizer_paras=opt_paras, verbose=True)
>>> model.fit(data.X_train_scaled, data.y_train)
>>> pred = model.predict(data.X_test_scaled)
>>> print(pred)
array([1, 0, 1, 0, 1])

CLS_OBJ_LOSSES = ['CEL', 'HL', 'KLDL', 'BSL']

create_network(X, y) → Tuple[MlpNumpy, ObjectiveScaler]

evaluate(y_true, y_pred, list_metrics=('AS', 'RS'))

Return the list of performance metrics on the given test data and labels.

Parameters:

• y_true (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• y_pred (array-like of shape (n_samples,) or (n_samples, n_outputs)) – Predicted values for X.

• list_metrics (list, default=("AS", "RS")) – You can get metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

objective_function(solution=None)

Evaluates the fitness function for classification metric

Parameters:

solution (np.ndarray, default=None) –

Returns:

result – The fitness value

Return type:

float

score(X, y, method='AS')

Return the metric on the given test data and labels.

In multi-label classification, this is the subset accuracy which is a harsh metric since you require for each sample that each label set be correctly predicted.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True labels for X.

• method (str, default="AS") – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

result – The result of selected metric

Return type:

float

scores(X, y, list_methods=('AS', 'RS'))

Return the list of metrics on the given test data and labels.

In multi-label classification, this is the subset accuracy which is a harsh metric since you require for each sample that each label set be correctly predicted.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True labels for X.

• list_methods (list, default=("AS", "RS")) – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

set_fit_request(*, lb: bool | None | str = '$UNCHANGED$', obj_weights: bool | None | str = '$UNCHANGED$', save_population: bool | None | str = '$UNCHANGED$', ub: bool | None | str = '$UNCHANGED$') → MhaMlpClassifier

Request metadata passed to the fit method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to fit.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

• lb (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for lb parameter in fit.

• obj_weights (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for obj_weights parameter in fit.

• save_population (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for save_population parameter in fit.

• ub (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for ub parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, return_prob: bool | None | str = '$UNCHANGED$') → MhaMlpClassifier

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to predict.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

return_prob (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for return_prob parameter in predict.

Returns:

self – The updated object.

Return type:

object

set_score_request(*, method: bool | None | str = '$UNCHANGED$') → MhaMlpClassifier

Request metadata passed to the score method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to score.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

method (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for method parameter in score.

Returns:

self – The updated object.

Return type:

object

class metaperceptron.core.mha_mlp.MhaMlpRegressor(hidden_size=50, act1_name='tanh', act2_name='sigmoid', obj_name='MSE', optimizer='OriginalWOA', optimizer_paras=None, verbose=True)

Bases: BaseMhaMlp, RegressorMixin

Defines the general class of Metaheuristic-based MLP model for Regression problems that inherit the BaseMhaMlp and RegressorMixin classes.

Parameters:

• hidden_size (int, default=50) – The number of hidden nodes

• act1_name (str, default='tanh') – Activation function for the hidden layer. The supported activation functions are: [“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”, “silu”]

• act2_name (str, default='sigmoid') – Activation function for the hidden layer. The supported activation functions are: [“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”, “silu”]

• obj_name (str, default="MSE") – Current supported objective functions, please check it here: https://github.com/thieu1995/permetrics

• optimizer (str or instance of Optimizer class (from Mealpy library), default = "OriginalWOA") – The Metaheuristic Algorithm that use to solve the feature selection problem. Current supported list, please check it here: https://github.com/thieu1995/mealpy. If a custom optimizer is passed, make sure it is an instance of Optimizer class.

• optimizer_paras (None or dict of parameter, default=None) – The parameter for the optimizer object. If None, the default parameters of optimizer is used (defined in https://github.com/thieu1995/mealpy.) If dict is passed, make sure it has at least epoch and pop_size parameters.

• verbose (bool, default=True) – Whether to print progress messages to stdout.

Examples

>>> from metaperceptron import MhaMlpRegressor, Data
>>> from sklearn.datasets import make_regression
>>> X, y = make_regression(n_samples=200, random_state=1)
>>> data = Data(X, y)
>>> data.split_train_test(test_size=0.2, random_state=1)
>>> data.X_train_scaled, scaler = data.scale(data.X_train, method="MinMaxScaler")
>>> data.X_test_scaled = scaler.transform(data.X_test)
>>> opt_paras = {"name": "GA", "epoch": 10, "pop_size": 30}
>>> model = MhaMlpRegressor(hidden_size=15, act1_name="relu", act2_name="sigmoid",
>>>             obj_name="MSE", optimizer="BaseGA", optimizer_paras=opt_paras, verbose=True)
>>> model.fit(data.X_train_scaled, data.y_train)
>>> pred = model.predict(data.X_test_scaled)
>>> print(pred)

create_network(X, y) → Tuple[MlpNumpy, ObjectiveScaler]

Returns:

• network (MLP, an instance of MLP network)

• obj_scaler (ObjectiveScaler, the objective scaler that used to scale output)

evaluate(y_true, y_pred, list_metrics=('MSE', 'MAE'))

Return the list of performance metrics of the prediction.

Parameters:

• y_true (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• y_pred (array-like of shape (n_samples,) or (n_samples, n_outputs)) – Predicted values for X.

• list_metrics (list, default=("MSE", "MAE")) – You can get metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

objective_function(solution=None)

Evaluates the fitness function for regression metric

Parameters:

solution (np.ndarray, default=None) –

Returns:

result – The fitness value

Return type:

float

score(X, y, method='RMSE')

Return the metric of the prediction.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• method (str, default="RMSE") – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

result – The result of selected metric

Return type:

float

scores(X, y, list_methods=('MSE', 'MAE'))

Return the list of metrics of the prediction.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• list_methods (list, default=("MSE", "MAE")) – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

set_fit_request(*, lb: bool | None | str = '$UNCHANGED$', obj_weights: bool | None | str = '$UNCHANGED$', save_population: bool | None | str = '$UNCHANGED$', ub: bool | None | str = '$UNCHANGED$') → MhaMlpRegressor

Request metadata passed to the fit method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to fit.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

• lb (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for lb parameter in fit.

• obj_weights (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for obj_weights parameter in fit.

• save_population (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for save_population parameter in fit.

• ub (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for ub parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, return_prob: bool | None | str = '$UNCHANGED$') → MhaMlpRegressor

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to predict.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

return_prob (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for return_prob parameter in predict.

Returns:

self – The updated object.

Return type:

object

set_score_request(*, method: bool | None | str = '$UNCHANGED$') → MhaMlpRegressor

Request metadata passed to the score method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to score.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

method (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for method parameter in score.

Returns:

self – The updated object.

Return type:

object

metaperceptron.core.traditional_mlp module

class metaperceptron.core.traditional_mlp.MlpClassifier(hidden_size=50, act1_name='tanh', act2_name='sigmoid', obj_name='NLLL', max_epochs=1000, batch_size=32, optimizer='SGD', optimizer_paras=None, verbose=False, **kwargs)

Bases: BaseMlpTorch

Defines the class for traditional MLP network for Classification problems that inherit the BaseMlpTorch class

Parameters:

• hidden_size (int, default=50) – The hidden size of MLP network (This network only has single hidden layer).

• act1_name (str, defeault="tanh") – This is activation for hidden layer. The supported activation are: {“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “silu”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”}.

• act2_name (str, defeault="sigmoid") – This is activation for output layer. The supported activation are: {“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “silu”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”}.

• obj_name (str, default="NLLL") – The name of objective for classification problem (binary and multi-class classification)

• max_epochs (int, default=1000) – Maximum number of epochs / iterations / generations

• batch_size (int, default=32) – The batch size

• optimizer (str, default = "SGD") – The gradient-based optimizer from Pytorch. List of supported optimizer is: [“Adadelta”, “Adagrad”, “Adam”, “Adamax”, “AdamW”, “ASGD”, “LBFGS”, “NAdam”, “RAdam”, “RMSprop”, “Rprop”, “SGD”]

• optimizer_paras (dict or None, default=None) – The dictionary parameters of the selected optimizer.

• verbose (bool, default=True) – Whether to print progress messages to stdout.

Examples

>>> from metaperceptron import MlpClassifier, Data
>>> from sklearn.datasets import make_regression
>>>
>>> ## Make dataset
>>> X, y = make_regression(n_samples=200, n_features=10, random_state=1)
>>> ## Load data object
>>> data = Data(X, y)
>>> ## Split train and test
>>> data.split_train_test(test_size=0.2, random_state=1, inplace=True)
>>> ## Scale dataset
>>> data.X_train, scaler = data.scale(data.X_train, scaling_methods=("minmax"))
>>> data.X_test = scaler.transform(data.X_test)
>>> ## Create model
>>> model = MlpClassifier(hidden_size=25, act1_name="tanh", act2_name="sigmoid", obj_name="BCEL",
>>>                 max_epochs=10, batch_size=32, optimizer="SGD", optimizer_paras=None, verbose=True)
>>> ## Train the model
>>> model.fit(data.X_train, data.y_train)
>>> ## Test the model
>>> y_pred = model.predict(data.X_test)
>>> ## Calculate some metrics
>>> print(model.score(X=data.X_test, y=data.y_test, method="RMSE"))
>>> print(model.scores(X=data.X_test, y=data.y_test, list_methods=["R2", "NSE", "MAPE"]))
>>> print(model.evaluate(y_true=data.y_test, y_pred=y_pred, list_metrics=["R2", "NSE", "MAPE", "NNSE"]))

CLS_OBJ_BINARY_1 = ['PNLLL', 'HEL', 'BCEL', 'CEL', 'BCELL']

CLS_OBJ_BINARY_2 = ['NLLL']

CLS_OBJ_LOSSES = ['CEL', 'HEL', 'KLDL']

CLS_OBJ_MULTI = ['NLLL', 'CEL']

SUPPORTED_LOSSES = {'BCEL': <class 'torch.nn.modules.loss.BCELoss'>, 'BCELL': <class 'torch.nn.modules.loss.BCEWithLogitsLoss'>, 'CEL': <class 'torch.nn.modules.loss.CrossEntropyLoss'>, 'GNLLL': <class 'torch.nn.modules.loss.GaussianNLLLoss'>, 'HEL': <class 'torch.nn.modules.loss.HingeEmbeddingLoss'>, 'KLDL': <class 'torch.nn.modules.loss.KLDivLoss'>, 'NLLL': <class 'torch.nn.modules.loss.NLLLoss'>, 'PNLLL': <class 'torch.nn.modules.loss.PoissonNLLLoss'>}

create_network(X, y) → Tuple[NeuralNetClassifier, ObjectiveScaler]

Returns:

• network (MLP, an instance of MLP network)

• obj_scaler (ObjectiveScaler, the objective scaler that used to scale output)

evaluate(y_true, y_pred, list_metrics=('AS', 'RS'))

Return the list of performance metrics on the given test data and labels.

Parameters:

• y_true (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• y_pred (array-like of shape (n_samples,) or (n_samples, n_outputs)) – Predicted values for X.

• list_metrics (list, default=("AS", "RS")) – You can get metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

fit(X, y)

score(X, y, method='AS')

Return the metric on the given test data and labels.

In multi-label classification, this is the subset accuracy which is a harsh metric since you require for each sample that each label set be correctly predicted.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True labels for X.

• method (str, default="AS") – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

result – The result of selected metric

Return type:

float

scores(X, y, list_methods=('AS', 'RS'))

Return the list of metrics on the given test data and labels.

In multi-label classification, this is the subset accuracy which is a harsh metric since you require for each sample that each label set be correctly predicted.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True labels for X.

• list_methods (list, default=("AS", "RS")) – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

set_predict_request(*, return_prob: bool | None | str = '$UNCHANGED$') → MlpClassifier

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to predict.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

return_prob (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for return_prob parameter in predict.

Returns:

self – The updated object.

Return type:

object

set_score_request(*, method: bool | None | str = '$UNCHANGED$') → MlpClassifier

Request metadata passed to the score method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to score.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

method (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for method parameter in score.

Returns:

self – The updated object.

Return type:

object

class metaperceptron.core.traditional_mlp.MlpRegressor(hidden_size=50, act1_name='tanh', act2_name='sigmoid', obj_name='MSE', max_epochs=1000, batch_size=32, optimizer='SGD', optimizer_paras=None, verbose=False, **kwargs)

Bases: BaseMlpTorch

Defines the class for traditional MLP network for Regression problems that inherit the BaseMlpTorch and RegressorMixin classes.

Parameters:

• hidden_size (int, default=50) – The hidden size of MLP network (This network only has single hidden layer).

• act1_name (str, defeault="tanh") – This is activation for hidden layer. The supported activation are: {“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “silu”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”}.

• act2_name (str, defeault="sigmoid") – This is activation for output layer. The supported activation are: {“none”, “relu”, “leaky_relu”, “celu”, “prelu”, “gelu”, “elu”, “selu”, “rrelu”, “tanh”, “hard_tanh”, “sigmoid”, “hard_sigmoid”, “log_sigmoid”, “silu”, “swish”, “hard_swish”, “soft_plus”, “mish”, “soft_sign”, “tanh_shrink”, “soft_shrink”, “hard_shrink”, “softmin”, “softmax”, “log_softmax”}.

• obj_name (str, default="MSE") – The name of loss function for the network.

• max_epochs (int, default=1000) – Maximum number of epochs / iterations / generations

• batch_size (int, default=32) – The batch size

• optimizer (str, default = "SGD") – The gradient-based optimizer from Pytorch. List of supported optimizer is: [“Adadelta”, “Adagrad”, “Adam”, “Adamax”, “AdamW”, “ASGD”, “LBFGS”, “NAdam”, “RAdam”, “RMSprop”, “Rprop”, “SGD”]

• optimizer_paras (dict or None, default=None) – The dictionary parameters of the selected optimizer.

• verbose (bool, default=True) – Whether to print progress messages to stdout.

Examples

>>> from metaperceptron import MlpRegressor, Data
>>> from sklearn.datasets import make_regression
>>>
>>> ## Make dataset
>>> X, y = make_regression(n_samples=200, n_features=10, random_state=1)
>>> ## Load data object
>>> data = Data(X, y)
>>> ## Split train and test
>>> data.split_train_test(test_size=0.2, random_state=1, inplace=True)
>>> ## Scale dataset
>>> data.X_train, scaler = data.scale(data.X_train, scaling_methods=("minmax"))
>>> data.X_test = scaler.transform(data.X_test)
>>> ## Create model
>>> model = MlpRegressor(hidden_size=25, act1_name="tanh", act2_name="sigmoid", obj_name="MSE",
>>>             max_epochs=10, batch_size=32, optimizer="SGD", optimizer_paras=None, verbose=True)
>>> ## Train the model
>>> model.fit(data.X_train, data.y_train)
>>> ## Test the model
>>> y_pred = model.predict(data.X_test)
>>> ## Calculate some metrics
>>> print(model.score(X=data.X_test, y=data.y_test, method="RMSE"))
>>> print(model.scores(X=data.X_test, y=data.y_test, list_methods=["R2", "NSE", "MAPE"]))
>>> print(model.evaluate(y_true=data.y_test, y_pred=y_pred, list_metrics=["R2", "NSE", "MAPE", "NNSE"]))

SUPPORTED_LOSSES = {'MAE': <class 'torch.nn.modules.loss.L1Loss'>, 'MSE': <class 'torch.nn.modules.loss.MSELoss'>}

create_network(X, y)

Returns:

• network (MLP, an instance of MLP network)

• obj_scaler (ObjectiveScaler, the objective scaler that used to scale output)

evaluate(y_true, y_pred, list_metrics=('MSE', 'MAE'))

Return the list of performance metrics of the prediction.

Parameters:

• y_true (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• y_pred (array-like of shape (n_samples,) or (n_samples, n_outputs)) – Predicted values for X.

• list_metrics (list, default=("MSE", "MAE")) – You can get metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

fit(X, y)

score(X, y, method='RMSE')

Return the metric of the prediction.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• method (str, default="RMSE") – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

result – The result of selected metric

Return type:

float

scores(X, y, list_methods=('MSE', 'MAE'))

Return the list of metrics of the prediction.

Parameters:

• X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.

• y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.

• list_methods (list, default=("MSE", "MAE")) – You can get all metrics from Permetrics library: https://github.com/thieu1995/permetrics

Returns:

results – The results of the list metrics

Return type:

dict

set_predict_request(*, return_prob: bool | None | str = '$UNCHANGED$') → MlpRegressor

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to predict.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

return_prob (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for return_prob parameter in predict.

Returns:

self – The updated object.

Return type:

object

set_score_request(*, method: bool | None | str = '$UNCHANGED$') → MlpRegressor

Request metadata passed to the score method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to score if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to score.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

New in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters:

method (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for method parameter in score.

Returns:

self – The updated object.

Return type:

object