Using Stratified Sampling to Improve LIME Image Explanations
This work addresses a specific bottleneck in explainable AI for computer vision, offering an incremental improvement to LIME.
The paper tackled artifacts in LIME Image explanations caused by Monte Carlo sampling undersampling, proposing a stratified sampling approach to improve explanation quality. Experiments demonstrated the method's efficacy.
We investigate the use of a stratified sampling approach for LIME Image, a popular model-agnostic explainable AI method for computer vision tasks, in order to reduce the artifacts generated by typical Monte Carlo sampling. Such artifacts are due to the undersampling of the dependent variable in the synthetic neighborhood around the image being explained, which may result in inadequate explanations due to the impossibility of fitting a linear regressor on the sampled data. We then highlight a connection with the Shapley theory, where similar arguments about undersampling and sample relevance were suggested in the past. We derive all the formulas and adjustment factors required for an unbiased stratified sampling estimator. Experiments show the efficacy of the proposed approach.