These terms are used for many meta-heuristic techniques, not just for PSO. They are also used in machine learning.
"Exploration" means ensuring that you examine lots of regions of the search space, so that you have less chance of getting trapped in a local minimum. "Exploitation" means that, if you find a promising region of the search space, you examine it carefully to see if it contains a good local minimum. These two things are to some extent in conflict, since, if you spend a lot of time on one, you have less time for the other.
Sometimes they are called "diversification" and "intensification", respectively.
thank you very much sir, but if we search in lots of region in search space we can get which is globally optimal but how it is possible that easily to get trapped for local optimum?? and what is the clear-cut difference between exploration and exploitation?
See, e.g., Section 5 of this paper:
http://kursinfo.himolde.no/forskningsgrupper/optimering/phdkurs/Metaheuristics%20in%20Combinatorial%20Optimization.pdf
Thank you very much sir the reference what you sent seem to be an ocean for me and it will take time for me once again thank you
Dear sir, DO you have references for Particle Swarm Optimization(PSO) and Improved PSO algorithms. mainly IPSO algorithm and also MATLAB programs for those techniques.?If so please send me sir, It will be very useful for my project..............
Sorry, heuristics is not really my area. (I work mostly on exact algorithms, based on integer programming.)
Exploitation means "utilizing the solution points in hand" and exploration means "searching for the unknown solution points". During exploitation, the search algorithm brings forth the solution found in the vicinity. Whereas, during exploration the search space is expanded to incorporate more search points. Actually heuristic algorithms work through amalgamating "EXPLOITATION" and "EXPLORATION", thereby improving the quality of solution.
For example in Genetic Algorithms (GA), crossover does the work of "exploitation" whereas "exploration" is rendered by mutation. Though this explanation is an introductory one, but would be helpful in en lighting your knowledge regarding Heuristic Algorithms.
Hi, Dear Vasim Akram
The particle swarm algorithm has shown ability to optimize in
continuous problem spaces, although it can struggle in problem spaces containing multiple optima. A variant, called Waves of Swarm Particles (WoSP),
has been shown to be able to handle problem spaces containing multiple optima by sequentially exploring these optima.
You can search for this paper : Quantised Problem Spaces and the
Particle Swarm Algorithm (By Tim Hendtlass)
I hope it will be useful.
For improved PSO algorithm in your field you can read some papers about NVPSO (A new vector particle swarm optimization) is proposed to solve constrained optimization problems.In this algorithm one dimensional search optimization methods are selected to produce a new position which is guaranteed to be feasible region for the particle which escapes from the feasible region.
Hello !!
In meta-heuristic optimization context, EXPLORATION is the method through which population based search algo tries to find as much wide search space as possible to avoid the local minimization (or maximization) and in case of EXPLOITATION, the algo tries to improve best found solution through various targeted approaches...
While designing a Metaheuristic, two contradictory criteria must be taken into account:
- Exploration of the search space that is known as diversification and
- Exploitation of the best solutions found that is known as intensification.
Promising regions are determined by the obtained “good” solutions.
- In intensification, the promising regions are explored more thoroughly in the hope to find better solutions.
- In diversification, nonexplored regions must be visited to be sure that all regions of the search space are evenly explored and that the search is not confined to only a reduced number of regions.
Exploitation: is to exploit what you already have as solutions. That's mean to modify some features of the actual solutions, trying to get better ones. This is what we call exploration or local search.
Exploration: is to "explore" the search space which means too simply to visit, new areas, regions, in the search space. More or less far of those already visited. This is also what we call global search. By exploration we avoid stagnation on a local optimum. Effectively the purpose is to find the global optimum.
EXPLOITATION or intensification: search for better solutions in the neighborhood of the current solution.
EXPLORATION or diversification: make sure the algorithm can explore the search
space efficiently.
Kindly, go through the following publication.
Y. del Valle, G. Venayagamoorthy, S. Mohagheghi, J.-C. Hernandez, R. Harley, Particle swarm optimization: Basic concepts, variants and applications in power systems, IEEE Transactions on Evolutionary Computation, 12 (2) (2008) 171-195.
I hope it will be very helpful to you.
M.N.Alam
https://www.researchgate.net/publication/280244004_A_comparative_study_of_metaheuristic_optimization_approaches_for_directional_overcurrent_relays_coordination
Article A comparative study of metaheuristic optimization approaches...
https://www.researchgate.net/publication/297245624_Particle_Swarm_Optimization_Algorithm_and_its_Codes_in_MATLAB
Research Particle Swarm Optimization: Algorithm and its Codes in MATLAB
https://www.researchgate.net/publication/296636431_Codes_in_MATLAB_for_Particle_Swarm_Optimization
Code Codes in MATLAB for Particle Swarm Optimization
the exploitation ( intensification ) phase task is browsing bounding region of the research domain based on one-hope connection of improving promising way (K). This operation amounts to intensifying the search in the surrounding of K.
in contrast, the exploration ( diversification) task explores palpating a much bigger region of the search domain based on one-hope connection technique for discovering the best solutions that are not yet been elaborated.