On Locality of Local Explanation Models
This work addresses interpretability issues in machine learning models for practitioners needing accurate local explanations, though it is incremental as it builds on existing Shapley methods.
The paper tackled the problem of misleading feature attributions from global Shapley values by proposing neighbourhood reference distributions to improve local interpretability, resulting in more meaningful sparse attributions and increased robustness to adversarial classifiers.
Shapley values provide model agnostic feature attributions for model outcome at a particular instance by simulating feature absence under a global population distribution. The use of a global population can lead to potentially misleading results when local model behaviour is of interest. Hence we consider the formulation of neighbourhood reference distributions that improve the local interpretability of Shapley values. By doing so, we find that the Nadaraya-Watson estimator, a well-studied kernel regressor, can be expressed as a self-normalised importance sampling estimator. Empirically, we observe that Neighbourhood Shapley values identify meaningful sparse feature relevance attributions that provide insight into local model behaviour, complimenting conventional Shapley analysis. They also increase on-manifold explainability and robustness to the construction of adversarial classifiers.