This article shows how to use performance and data profile benchmarking tools to improve the performance of algorithms. We propose to achieve this goal by defining and approximately solving suitable optimization problems involving the parameters of the algorithm under consideration. Because these problems do not have derivatives and may involve integer variables, we suggest using a mixed-integer derivative-free optimizer for this task. A numerical illustration is presented (using the BFO package), which indicates that the obtained gains are potentially significant.

A note on using performance and data profiles for training algorithms / Porcelli M.; Toint P. L.. - In: ACM TRANSACTIONS ON MATHEMATICAL SOFTWARE. - ISSN 0098-3500. - ELETTRONICO. - 45:(2019), pp. 20.1-20.10. [10.1145/3310362]

A note on using performance and data profiles for training algorithms

Porcelli M.;
2019

Abstract

This article shows how to use performance and data profile benchmarking tools to improve the performance of algorithms. We propose to achieve this goal by defining and approximately solving suitable optimization problems involving the parameters of the algorithm under consideration. Because these problems do not have derivatives and may involve integer variables, we suggest using a mixed-integer derivative-free optimizer for this task. A numerical illustration is presented (using the BFO package), which indicates that the obtained gains are potentially significant.
2019
45
1
10
Porcelli M.; Toint P. L.
File in questo prodotto:
File Dimensione Formato  
pt4_perfprof_ACM_2019_def.pdf

Accesso chiuso

Licenza: Tutti i diritti riservati
Dimensione 464.07 kB
Formato Adobe PDF
464.07 kB Adobe PDF   Richiedi una copia
2019_toms_pt3.pdf

Accesso chiuso

Licenza: Tutti i diritti riservati
Dimensione 699.17 kB
Formato Adobe PDF
699.17 kB Adobe PDF   Richiedi una copia

I documenti in FLORE sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificatore per citare o creare un link a questa risorsa: https://hdl.handle.net/2158/1351280
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 5
  • ???jsp.display-item.citation.isi??? 5
social impact