Table 1.

Algorithm: LIME.

Algorithm: LIME
Input: (1) Complex Model f; (2) Samples X; (3) Number of randomly generated samples N
Steps: 1. Through feature screening, the more important d′ features are preliminarily obtained, allowing the interpretation version X′ of X to be obtained 2. A new sample Z′ is generated by randomly perturbing X′; then, Z′ is restored to Z with the same dimensions as X. The complex model is used to predict and obtain the labels 3. The newly generated dataset is fitted with a linear model
Output: The weight of the linear model