teoshibin / COMP2024_AIM_CW

AI method coursework BBOB optimization MATLAB code and datasets

Function Optimization Benchmarking

AI method coursework function optimization algorithm MATLAB code and datasets

Usage

Benchmarking

• uncomment and comment different algorithms in exampleexperiment

• generate data with 2010 BBOB framework by running exampleexperiment or dedicated experiment script in MATLAB

  1. set path that points to fgeneric as "pwd" e.g. addpath(pwd)
  2. set fgeneric delta precision = 1e-16
  3. set datapath = '../Datasets/ALGONAME'
  4. set maxfunevals = '5e3'
  5. set random generator seed rng('default')

• post-process data with numbbo framework

   	python -m cocopp [-o OUTPUT_FOLDERNAME] YOURDATAFOLDER [MORE_DATAFOLDERS]

  Example command

   	cd Datasets
   	python -m cocopp -o "..\Post_Processing\PSO_ppdata" "PSO"

  Note
  Path to datasets will be used as algorithm name in ppdata, this is why we cd into the datasets

Generate Excel Score

see info2excel documentation

Results

Post Processing

Excel Scores

All tested algorithm with their corresponding scores in 5d

Algo	Score 5d	In Report
CMAES	2344	0
DEAE	2322	1
JADEb	1941	0
JADE	1868	1
JADEctpb	1868	0
DE	1083	1
DEb	1083	0
PSO_Bounds	500	1
PSO	234	1
Bin_GA	-14	1
PSO_EDA	-49	1