In this video we will cover a numerical technique for estimating Shapley Values. Originally developed within Game Theory, the approach we will take here is far more applicable within the context of machine learning, when compared with the classic definition. Note the method covered here was originally published by Strumbelj et al. 2014 (see: link.springer.com/article/10.1007/s10115-013-0679-x).
The break-down of the video is as follows:
Introduction 00:00
Motivation 00:43
Two Issues with the Classic Definition 01:26
Algorithm Description 03:28
Jupyter Notebook Coding Example 12:07
Conclusions 21:15
The best way to keep up-to-date with my video/blog content is to sign up for my monthly Newsletter! Please visit: insidelearningmachines.com/newsletter/ to register.
The notebook presented here can be found at: github.com/insidelearningmachines/Blog/blob/main/Notebook%20XXXIII%20Shapley%20Values%20for%20Machine%20Learning.ipynb
This video is based off of an article on my blog. You can find that blog article here: insidelearningmachines.com/shapley_values/
The homepage of my blog is: insidelearningmachines.com
Other social media includes:
Twitter: inside_machines
Facebook: Inside-Learning-Machines-112215488183517
#machinelearning #datascience #shapley #xai #insidelearningmachines