DiCE
Diverse Counterfactual Explanations (DiCE) is another popular XAI framework that we briefly covered in Chapter 2, Model Explainability Methods, for the CFE tutorial. Interestingly, DiCE is also one of the key XAI frameworks from Microsoft Research, but it is yet to be integrated with the InterpretML module (I wonder why!). I find the entire idea of CFE to be very close to the ideal human-friendly explanation that gives actionable recommendations. This blog from Microsoft discusses the motivation and idea behind the DiCE framework: https://www.microsoft.com/en-us/research/blog/open-source-library-provides-explanation-for-machine-learning-through-diverse-counterfactuals/.
In comparison to ALIBI CFE, I found DiCE to produce more appropriate CFEs with minimal hyperparameter tuning. That's why I feel it's important to mention DiCE, as it is primarily designed for example-based explanations. Next, let's discuss the CFE methods that are supported in DiCE.