Abstract | ||
---|---|---|
The ACO(mathbb {_R}) algorithm is based on the Ant Colony Optimization (ACO) metaphor, and a crossover operator does not naturally within this metaphor. In spite of this, we investigate in this paper whether the performance of ACO(mathbb {_R}) would benefit from the deployment, with a fixed probability, of a crossover operator. Our extensive experimental evaluation uses two applications: (1) training feedforward neural networks for classification using 65 benchmark datasets from the UCI repository; and (2) optimizing several popular synthetic benchmark continuous-domain functions with the number of dimensions varying from 10 up to 10,000. Our experimental results confirm that the use of crossover does improve performance on both applications to a statistically significant extent. |
Year | Venue | Field |
---|---|---|
2018 | ANTS Conference | Ant colony optimization algorithms,Feedforward neural network,Crossover,Computer science,Algorithm,Operator (computer programming) |
DocType | Citations | PageRank |
Conference | 0 | 0.34 |
References | Authors | |
16 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ashraf M. Abdelbar | 1 | 243 | 25.43 |
Khalid M. Salama | 2 | 160 | 13.09 |