Skip to content

Latest commit

 

History

History
14 lines (10 loc) · 903 Bytes

README.md

File metadata and controls

14 lines (10 loc) · 903 Bytes

WCET_via_EVT

The worst case execution times via extreme value theory

This is the repository for the source code and experiments for Worst-Case Convergence Time of ML Algorithms via Extreme Value Theory, which has been accepted to the 3'rd Conference on AI Engineering (CAIN'24).

To run the extreme value distributions over SnapBuddy example:

Rscript extreme_analysis.R Dataset/Snapbuddy_timing.txt 15.6

You can use any other dataset of experiments and the threshold can be found in the paper for each experiment.

Note that jupyter notebook for the extreme value theory is included.

Please feel free to see the source code for programs and cyber-phyiscal DNN controllers. The source code for CPS include jupyter notebook that illustrtates the steps in interacting and analyzing the controler to reach to an equilibrium state.