Using Stratified Sampling to Improve LIME Image Explanations
Date:
Poster Presentation at AAAI-24
Using Stratified Sampling to Improve LIME Image Explanations was presented at the Thirty-Eighth AAAI Conference on Artificial Intelligence (AAAI-24) in Vancouver, Canada, on February 24, 2024.
The paper was accepted to the AAAI-24 Main Technical Track – Machine Learning and published in the Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 38, pp. 14785–14792.
The work was presented during the conference as a poster presentation. AAAI-24 organized poster sessions for accepted technical papers, while a comparatively limited subset of papers was additionally assigned oral presentation slots. Therefore, an oral presentation should not be interpreted as the criterion for acceptance into the AAAI technical program; rather, it represented an additional presentation format for selected papers.
The poster session provided an opportunity to present our approach for improving the reliability of LIME image explanations through stratified sampling and to discuss the methodology and results directly with researchers attending AAAI-24.
A recorded presentation is also available through the official AAAI-24 collection on Underline.
Poster Presentation at AAAI-24 in Vancouver
Authors: Muhammad Rashid, Elvio G. Amparore, Enrico Ferrari, and Damiano Verda