Consistent and Truthful Interpretation with Fourier Analysis
share
For many interdisciplinary fields, ML interpretations need to be consistent with what-if scenarios related to the current case, i.e., if one factor changes, how does the model react? Although the attribution methods are supported by the elegant axiomatic systems, they mainly focus on individual inputs, and are generally inconsistent. To support what-if scenarios, we introduce a new notion called truthful interpretation, and apply Fourier analysis of Boolean functions to get rigorous guarantees. Experimental results show that for neighborhoods with various radii, our method achieves 2x - 50x lower interpretation error compared with the other methods.
READ FULL TEXT