SpiderMonkey.jl

A Julia package for Spider Monkey Optimization.
Author akaysh
Popularity
9 Stars
Updated Last
2 Years Ago
Started In
April 2020

SpiderMonkey.jl

Build Status Coverage Status

Spider Monkey Optimization is a numerical optimization algorithm which is based on the foraging begavior of spider monkeys. It is one of the swarm intelligence approaches which can be broadly classified as an algorithm inspired by intelligent foraging behavior of fission–fusion social structure based animals. The animals which follow fission–fusion social systems (like Spider Monkeys), split themselves from large to smaller groups and vice-versa based on the scarcity or availability of food.
Here is the research paper for SMO algorithm and here is my Aegist Spider Monkey algorithm research paper which proposed a few tweaks in the original algorithm to make it more efficient (will be implemented soon as a part of this package).

Install

Pkg.add("SpiderMonkey")

Usage

Import the library

This brings the function smo (short for Spider Monkey Optimization) into the scope.

using SpiderMonkey

Define your optimization function and parameters

The function can be defined like the ones in BlackBoxOptimizationBenchmarking or you can use one of those implemented bechmarking functions.

For the SMO optimization algorithm we need the function f to be optimized, the dimension of the search space D or the function and population of the spider monkeys P for the heuristic search. We also need to define the upper ub and lower bounds lb of the search space coordinates for the boundaries which are DX1 arrays.

SMO

using BlackBoxOptimizationBenchmarking;
ub = fill(5,(100,));
lb = fill(-5,(100,));
smo(100,30,BlackBoxOptimizationBenchmarking.F1.f,0.5,25,20,8,100,ub,lb);

Best hyperparameters?

There are lots of parameters that can be tweaked. The algorithm is generally fairly robust meaning that slight changes in parameters should not result in drastically different performance.

I suggest reading on Spider Monkey Optimization algorithm here before starting. However if you don't feel like it, the parameters worth playing around with might be:

  • ɳ: perturbation rate
  • λ: local leader limit, max count for which the position of the local leader remains unchanged
  • α: global leader limit, max count for which the position of the global leader remains unchanged
  • β: maximum number of groups allowed
  • τ: maximum number of iterations

Used By Packages

No packages found.