Can I Trust the Explanations? Investigating Explainable Machine Learning Methods for Monotonic Models
This work addresses the reliability of explanations for domain-specific monotonic models, which is incremental as it builds on existing explainable methods.
The paper investigates whether explainable machine learning methods provide consistent scientific explanations when applied to science-informed monotonic models, finding that Shapley values work well for individual monotonicity, while Integrated gradients are reasonable for strong pairwise monotonicity on average.
In recent years, explainable machine learning methods have been very successful. Despite their success, most explainable machine learning methods are applied to black-box models without any domain knowledge. By incorporating domain knowledge, science-informed machine learning models have demonstrated better generalization and interpretation. But do we obtain consistent scientific explanations if we apply explainable machine learning methods to science-informed machine learning models? This question is addressed in the context of monotonic models that exhibit three different types of monotonicity. To demonstrate monotonicity, we propose three axioms. Accordingly, this study shows that when only individual monotonicity is involved, the baseline Shapley value provides good explanations; however, when strong pairwise monotonicity is involved, the Integrated gradients method provides reasonable explanations on average.