A Learning Theoretic Perspective on Local Explainability
This work addresses the theoretical underpinnings of local explainability in machine learning, which is important for researchers and practitioners seeking reliable interpretability, though it appears incremental in connecting existing concepts.
The paper tackles the problem of linking interpretable machine learning with learning theory by analyzing local approximation explanations, showing that a model's test-time accuracy can be bounded based on its local explainability and addressing explanation generalization for finite sample-based methods, with empirical validation.
In this paper, we explore connections between interpretable machine learning and learning theory through the lens of local approximation explanations. First, we tackle the traditional problem of performance generalization and bound the test-time accuracy of a model using a notion of how locally explainable it is. Second, we explore the novel problem of explanation generalization which is an important concern for a growing class of finite sample-based local approximation explanations. Finally, we validate our theoretical results empirically and show that they reflect what can be seen in practice.