In a direct search method, algorithms are proposed to solve optimization problems. After developing the algorithm it is important to test the performance, effectiveness and speed of the algorithm. My question is how to compute all these and compare the results with some other algorithm
Dear Nabajyoti Bhattacharje,
I suggest you to see links and attached files on topic.
https://ieeexplore.ieee.org/document/7105570
https://hal.archives-ouvertes.fr/hal-02571791/document
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0122827
Conference Paper Applying particle swarm optimization to software testing
Article Performance evaluation of the particle swarm optimization al...
Best regards
Hello dear Nabajyoti Bhattacharjee
Performance analysis of a swarm algorithm is a necessary task to perform and must be done on a fair basis. A theoretical approach is generally not sufficient to evaluate a swarm algorithm. To addresses some guidelines of evaluating experimentally a swarm algorithm and/or comparing metaheuristics in a rigorous way. To evaluate the performance of a swarm algorithm in a rigorous manner, the following three steps must be considered:
1) Experimental design: In the first step, the goals of the experiments, the selected
instances, and factors have to be defined.
2) Measurement: In the second step, the measures to compute are selected. After
executing the different experiments, statistical analysis is applied to the obtained
results. The performance analysis must be done with state-of-the-art optimization
algorithms dedicated to the problem.
3) Reporting: Finally, the results are presented in a comprehensive way, and an
analysis is carried out following the defined goals. Another main issue here is to
ensure the reproducibility of the computational experiments.
T. Bartz-Beielstein. Experimental Research in Evolutionary Computation. Springer, 2006.
I hope; I made sense.
Solving the test problems like Ackley, Rosenbrock, Sphere etc to evaluate the effectiveness and speed of algorithm and on comparing the results with the established results of different standard algorithms may be one of the way to test the efficiency of developed algorithm
Indeed, it's based on the problem space that you need to solve whether (Continuous or Combinatorial/discrete). All these Functions have mentioned like Ackley, Rosenbrock, Sphere it's continuous problem.
Subsequently, if you proposed a new algorithm or you have a modification in any algorithm, of course, you need to use many of these Function problems to test the efficiency of the developed algorithm.
In case your problem based on combinatorial/discrete you need to use the benchmark datasets of the problem like vehicle routing problem, travel seaman problem, time tableting problem, and so on to test the efficiency of the developed algorithm.
Regards
Dear Nabajyoti Bhattacharjee
Kindly check the attached links and files.
Article A Comprehensive Review of Swarm Optimization Algorithms
https://ieeexplore.ieee.org/abstract/document/6426820
https://ieeexplore.ieee.org/abstract/document/6726474
Article Applications of Particle Swarm Optimization in Geotechnical ...
The convergence rate and statistical tests results are very popular in order to evaluate and compare the performance of swarm or evolutionary algorithm. In the following two relative works are
Article New Insights into Position optimisation of Wave Energy Conve...
Article A Hybrid Cooperative Co-evolution Algorithm Framework for Op...
- Badges
- Science topic