On July 9th 2024, Isabelle Puaut and Hugo Reymond presented their work at the 22nd International Workshop on Worst-Case Execution Time Analysis (WCET 2024) during the ECRTS conference.
Isabelle Puaut gave a keynote about her experience using ML for WCET calculation. Rather than presenting only what worked, Isabelle also discussed in this keynote the bad, and even very bad, surprises encountered during the process, and how to overcome (most of) them.
You can find her publications on the subject on her personal page.
Hugo Reymond and Abderaouf Nassim Amalou presented WORTEX : Worst Case Execution Time and Energy Consumption Estimation using Explainable AI.
This work, in collaboration with Hector CHABOT and Isabelle Puaut, aims at providing a way to build WCET/WCEC model when micro-architecture details are not available, and better understand the reasoning of ML-based WCET estimation models to be able to debug them.
If you want to know more about it, the paper will soon be available in its final version, but a preliminary one is available : https://submission.dagstuhl.de/collections/WCET-2024/preliminary-proceedings/1
Also, the dataset created for this paper is under creative common licensing, so feel free to have a look at it ! https://zenodo.org/records/11066623
