Counterfactual Explanations for Explainable AI

Explainable models in artificial intelligence (AI) have become necessary for the business world. It has been greatly accelerated by introducing global policies and increased awareness around data protection and privacy. Towards this, the area of explainable AI has become a valuable tool for explaining the black box decision of ML models. This whitepaper discusses the basic concepts of Counterfactual explanations, their utilities, and how it helps mitigate the bias in decisions made by an automated system.

In this whitepaper, you will learn:

• What is Counterfactual thinking?
• What are Counterfactual explanations in XAI?
• What are the properties of actionable CFEs?
• What are the business benefits of CFEs?
• Model fairness through CF generation