| Naslov: | Using machine learning methods to assess module performance contribution in modular optimization frameworks |
|---|
| Avtorji: | ID Kostovska, Ana, Institut "Jožef Stefan" (Avtor)
ID Vermetten, Diederick (Avtor)
ID Korošec, Peter, Institut "Jožef Stefan" (Avtor)
ID Džeroski, Sašo, Institut "Jožef Stefan" (Avtor)
ID Doerr, Carola (Avtor)
ID Eftimov, Tome, Institut "Jožef Stefan" (Avtor) |
|---|
| Datoteke: |  URL - Izvorni URL, za dostop obiščite https://direct.mit.edu/evco/article-abstract/doi/10.1162/evco_a_00356/123899/Using-Machine-Learning-Methods-to-Assess-Module?redirectedFrom=fulltext
 PDF - Predstavitvena datoteka, prenos (1,37 MB)
MD5: 7D7A574A260449EFC752BD8129275844
|
|---|
| Jezik: | Angleški jezik |
|---|
| Tipologija: | 1.01 - Izvirni znanstveni članek |
|---|
| Organizacija: |  IJS - Institut Jožef Stefan
|
|---|
| Povzetek: | Modular algorithm frameworks not only allow for combinations never tested in manually selected algorithm portfolios, but they also provide a structured approach to assess which algorithmic ideas are crucial for the observed performance of algorithms. In this study, we propose a methodology for analyzing the impact of the different modules on the overall performance. We consider modular frameworks for two widely used families of derivative-free black-box optimization algorithms, the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) and differential evolution (DE). More specifically, we use performance data of 324 modCMA-ES and 576 modDE algorithm variants (with each variant corresponding to a specific configuration of modules) obtained on the 24 BBOB problems for 6 different runtime budgets in 2 dimensions. Our analysis of these data reveals that the impact of individual modules on overall algorithm performance varies significantly. Notably, among the examined modules, the elitism module in CMA-ES and the linear population size reduction module in DE exhibit the most significant impact on performance. Furthermore, our exploratory data analysis of problem landscape data suggests that the most relevant landscape features remain consistent regardless of the configuration of individual modules, but the influence that these features have on regression accuracy varies. In addition, we apply classifiers that exploit feature importance with respect to the trained models for performance prediction and performance data, to predict the modular configurations of CMA-ES and DE algorithm variants. The results show that the predicted configurations do not exhibit a statistically significant difference in performance compared to the true configurations, with the percentage varying depending on the setup (from 49.1% to 95.5% for modCMA and 21.7% to 77.1% for DE) |
|---|
| Ključne besede: | evolutionary computation, modular algorithm frameworks, DE |
|---|
| Status publikacije: | Objavljeno |
|---|
| Verzija publikacije: | Recenzirani rokopis |
|---|
| Poslano v recenzijo: | 16.07.2023 |
|---|
| Datum sprejetja članka: | 03.07.2024 |
|---|
| Datum objave: | 17.10.2024 |
|---|
| Založnik: | Massachusetts Institute of Technology |
|---|
| Leto izida: | 2024 |
|---|
| Št. strani: | 31 str. |
|---|
| Izvor: | ZDA |
|---|
| PID: | 20.500.12556/DiRROS-20968  |
|---|
| UDK: | 004.85 |
|---|
| ISSN pri članku: | 1530-9304 |
|---|
| DOI: | 10.1162/evco_a_00356 |
|---|
| COBISS.SI-ID: | 206993155 |
|---|
| Avtorske pravice: | © 2024 Massachusetts Institute of Technology |
|---|
| Opomba: | Nasl. z nasl. zaslona;
Opis vira z dne 10. 9. 2024;
|
|---|
| Datum objave v DiRROS: | 11.12.2024 |
|---|
| Število ogledov: | 79 |
|---|
| Število prenosov: | 30 |
|---|
| Metapodatki: |    |
|---|
|
:
|
Kopiraj citat |
|---|
| | | | Objavi na: |  |
|---|
Postavite miškin kazalec na naslov za izpis povzetka. Klik na naslov izpiše
podrobnosti ali sproži prenos. |
