Scribd
Upload
3 views1 page

Convert and Merge These Two Algorithm Into Python Code and Create A New Algorithm

This document outlines a new algorithm that merges the LIME and KernelSHAP algorithms for local explanations in Python. The new algorithm takes a classifier, an input sample, and parameters for superpixels and feature selection, then combines steps from both algorithms to generate explainable coefficients. The final output is the coefficients from a linear model fitted on the selected features based on similarity scores and SHAP values.

Uploaded by

ZuniButt
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
3 views1 page

Convert and Merge These Two Algorithm Into Python Code and Create A New Algorithm

This document outlines a new algorithm that merges the LIME and KernelSHAP algorithms for local explanations in Python. The new algorithm takes a classifier, an input sample, and parameters for superpixels and feature selection, then combines steps from both algorithms to generate explainable coefficients. The final output is the coefficients from a linear model fitted on the selected features based on similarity scores and SHAP values.

Uploaded by

ZuniButt
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 1

Convert and merge these two algorithm into python code and create a new algorithm.

Algorithm 1 LIME algorithm for local explanations

Input: classifier f, input sample x, number of superpixels n, number of features to pick m

Output: explainable coefficients from the linear model

1: ŷ ← f.predict(x)

2: for i in n do

3: p_i ← Permute(x) //Randomly pick superpixels

4: obs_i ← f.predict(p)

5: dist_i ← |ŷ - obs_i|

6: end for

7: simscore ← SimilarityScore(dist)

8: x_pick ← Pick(psimscorem)

9: L ← LinearModel.fit(pmsimscore)

10: return L.weights

Algorithm 2 KernelSHAP Algorithm for Local Explanations


Input: classifier f, input sample x
Output: explainable coefficients from the linear model
1: zk ← SampleByRemovingFeature(x)
2: zk ← h(zk) // hx is a feature transformation to reshape to x
3: yk ← f(zk)
4: Wx ← SHAP(f, zk, yk)
5: LinearModel(Wx).fit()
6: Return LinearModel.coefficients()

You might also like