Marginal Contribution Feature Importance – an Axiomatic Approach for Explaining Data. Amnon Catav (TAU).

Marginal Contribution Feature Importance – an Axiomatic Approach for Explaining Data

Abstract:
In recent years, methods were proposed for assigning feature importance scores to measure the contribution of individual features. While in some cases the goal is to understand a specific model, in many cases the goal is to understand the contribution of certain properties (features) to a real-world phenomenon. Thus, a distinction has been made between feature importance scores that explain a model and scores that explain the data.

In this talk, we will demonstrate the cavities of existing SOTA methods for explaining the data and introduce MCI, a novel feature importance score that is based on an axiomatic approach.