sklearn_pmml_model.neural_network._base#
Module Contents#
Classes#
Abstract class for Neural Network models. |
- class sklearn_pmml_model.neural_network._base.PMMLBaseNeuralNetwork#
Abstract class for Neural Network models.
The PMML model consists out of a <NeuralNetwork> element, containing a <NeuralInputs> element that describes the input layer neurons with <NeuralInput> elements. Next, a <NeuralLayer> element describes all other neurons with associated weights and biases. The activation function is either specified globally with the activationFunction attribute on the <NeuralNetwork> element, or the same attribute on each layer. Note however that scikit-learn only supports a single activation function for all hidden layers. Finally, the <NeuralOutputs> element describes the output layer. The output is currently expected to match the target field in <MiningSchema>.
Notes
Specification: http://dmg.org/pmml/v4-3/NeuralNetwork.html