Explainability is a gateway between Artificial Intelligence and society as current popular deep learning models are generally weak in explaining the process and prediction results. Local Interpretable Model-agnostic (LIME) is a recent technique that explains the predictions of any faithfully by learning an interpretable model locally around the. However, the sampling operation in the standard implementation of is defective. Perturbed samples are generated from a uniform distribution, the complicated correlation between features. This paper proposes a Modified Perturbed Sampling operation for LIME (MPS-LIME), which is as the clique set construction problem. In image classification,-LIME converts the superpixel image into an undirected graph. Various show that the MPS-LIME explanation of the black-box model achieves better performance in terms of understandability, fidelity, and.
No takes yet. Share an insight, caveat, or question.
Shi et al. (2020) studied this question.