Example-based methods
Another approach to model explainability is provided by example-based methods. The idea of example-based methods is similar to how humans try to explain a new concept. As human beings, when we try to explain or introduce something new to someone else, often, we try to make use of examples that our audience can relate to. Similarly, example-based methods, in the context of XAI, try to select certain instances of the dataset to explain the behavior of the model. It assumes that observing similarities between the current instance of the data with a historic observation can be used to explain black-box models.
These are mostly model-agnostic approaches that can be applied to both structured and unstructured data. If the structured data is high-dimensional, it becomes slightly challenging for these approaches, and all the features cannot be included to explain the model. So, it works well only if there is an option to summarize the data instance or pick up only...
