What is it about?
Global Sensitivity Analaysis methods can give your information about which features are important for a particular target function or model by simply providing an existing data set or generating a design of experiments to be evaluated. In this paper we review a wide variety of these global sensitivity methods, and we give recommendations on which method to use in which situation. It is crucial to better understand your machine learning models or get an understanding of complex high dimensional real-world processes in order to improve or learn from them.
Featured Image
Photo by Maxime VALCARCE on Unsplash
Why is it important?
Explainable Artificial Intelligence (XAI) is crucial in order to develop and deploy AI methods in real-world environments. Global sensitivity analysis (GSA) quantifies the importance of model inputs and their interactions with respect to model output. It is therefore an important method in the XAI field to better understand complex machine learning models. In this work we compare a wide range of GSA methods, we find their pros and cons and we recommend which methods to use in different contexts.
Read the Original
This page is a summary of: A Comparison of Global Sensitivity Analysis Methods for Explainable AI with an Application in Genomic Prediction, IEEE Access, January 2022, Institute of Electrical & Electronics Engineers (IEEE),
DOI: 10.1109/access.2022.3210175.
You can read the full text:
Resources
Contributors
The following have contributed to this page
