Optimisation_AI_Cotonou2021

This is the project for the introduction to optimization class in Cotonou 2021

• Course notes are in Slides/cours_optim_ML_Cotonou2021.pdf.
• Project statement can be found in Slides/project.pdf.
• The Project directory contains R code for : gradient based descent algorithms (basic, with line search, with momentum and Nesterov acceleration), a restarted version of the descent algorithm, several test functions and associated visualizations. There is also a basic neural network whose training serves as an application. Everything is coded from scratch.

(July 2022) the momentum and NAG implementations may be slightly buggy (alpha bar terms forgotten). Refer to EEAI 2022 version for update in python.