Challenges in interpretability of additive models
By: Xinyu Zhang, Julien Martinelli, ST John
Potential Business Impact:
Makes computer brains easier to understand.
We review generalized additive models as a type of ``transparent'' model that has recently seen renewed interest in the deep learning community as neural additive models. We highlight multiple types of nonidentifiability in this model class and discuss challenges in interpretability, arguing for restraint when claiming ``interpretability'' or ``suitability for safety-critical applications'' of such models.
Similar Papers
Interpretable additive model for analyzing high-dimensional functional time series
Methodology
Predicts future trends from many related past events.
An Interpretable Deep Learning Model for General Insurance Pricing
Machine Learning (CS)
Explains insurance costs clearly, like a smart calculator.
Additive Models Explained: A Computational Complexity Approach
Machine Learning (CS)
Explains how computer models make decisions.