Escaping Local Optima Using Crossover with Emergent Diversity
- Submitting institution
-
The University of Sheffield
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 2490
- Type
- D - Journal article
- DOI
-
10.1109/TEVC.2017.2724201
- Title of journal
- IEEE Transactions on Evolutionary Computation
- Article number
- -
- First page
- 484
- Volume
- 22
- Issue
- 3
- ISSN
- 1089-778X
- Open access status
- Compliant
- Month of publication
- August
- Year of publication
- 2017
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
7
- Research group(s)
-
A - Algorithms
- Citation count
- 27
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- For the first time the benefits of evolving a population via crossover, mutation and selection in a standard genetic algorithm (GA) are rigorously proven. Shows how interplay of crossover and mutation leads to the emergence of diversity. This leads to linear speed-ups by escaping local optima compared to the use of one operator alone. Furthermore, optimal mutation rates for GAs are considerably larger than those recommended for mutation-only algorithms. Several works have built upon it, emphasising the benefits of higher mutation rates in other settings, (doi.org/10.1145/3071178.3071301, doi.org/10.1109/TEVC.2017.2745715, doi.org/10.1016/j.artint.2019.03.001). Conference version nominated for a best paper award at PPSN 2016.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -