Using LIME (Local Interpretable Model-agnostic Explanations) to explain image classification model. ResNet is used as the pre-trained image classification model that LIME explains. The aim is to provide clear, interpretable explanations for the predictions of image classifiers, enabling better understanding of what features influence a model's decisions. LIME generates human-readable explanations by highlighting the regions of the image that were most influential in the model's prediction.
- The image in the center highlights the top 5 superpixels that had the most positive influence on the model's classification decision (contributed the most to the prediction with the highest probability).
- The image on the right shows the top 5 superpixels that had the most negative impact on the model's classification decision (areas that detracted from or opposed the predicted class).