Train and Explain Keras Network with Counterfactuals

This application is a simple example of using Conterfactual Explanations (Python) Component to identify the counterfactual instances for a Binary classification model trained in KNIME with Keras.

The Component (in yellow) uses KNIME Python Integration for normalisation of features and outputs the features along with the pickled file containing the preprocessing object. The Component (in blue) can be used to select the instances to be used for Counterfactual Explainations.

External resources

• Counterfactual - Interpretable-ml-book
• How to set up KNIME Python extension
• KNIME Keras Integration
• Python Library for Counterfactual Instances