Stage-oe-small.jpg

MSc. Nugroho Fredivianus: Unterschied zwischen den Versionen

Aus Aifbportal
Wechseln zu:Navigation, Suche
 
(Eine dazwischenliegende Version von einem anderen Benutzer wird nicht angezeigt)
Zeile 4: Zeile 4:
 
|Beschreibung DE=Accuracy-based Learning Classifier System (XCS) is an online genetic-based machine learning method introduced by Wilson in 1995. It learns by maintaining several rules called classifiers in form of an input-output pair, where the input is a series of “genetic” information. XCS has the goals to learn the given task by responding to the environmental with a proper action and maximizing the awarded feedback. It also aims to generalize the rules in the population in order to reduce the usage of resources, along with an effort to minimize the time requirement.
 
|Beschreibung DE=Accuracy-based Learning Classifier System (XCS) is an online genetic-based machine learning method introduced by Wilson in 1995. It learns by maintaining several rules called classifiers in form of an input-output pair, where the input is a series of “genetic” information. XCS has the goals to learn the given task by responding to the environmental with a proper action and maximizing the awarded feedback. It also aims to generalize the rules in the population in order to reduce the usage of resources, along with an effort to minimize the time requirement.
  
XCS employs genetic operators (i.e., crossover and mutation) to discover general and accurate rules. However, these techniques have some drawbacks due to some randomizations (e.g., when picking crossover points) that potentially slow down the discovery process. A heuristic-based genetic operator called Rule Combining (RC) is introduced to address this issue. Instead of involving some random values, it generalizes rules by performing inductive reasoning on a pair of experienced classifiers and creates a child based upon it. Comparisons indicate that the RC technique is capable of improving the performance of XCS in fulfilling the given tasks. In particular, experiments indicate that it requires less learning time and a significantly smaller number of classifiers than the classical XCS.
+
XCS employs genetic operators (i.e., crossover and mutation) to discover general and accurate rules. However, these techniques have some drawbacks due to some randomizations (e.g., when picking crossover points) that potentially slow down the discovery process. A heuristic-based genetic operator called Rule Combining (RC) is introduced to address this issue. Instead of involving some random values, it generalizes rules by performing inductive reasoning on a pair of experienced classifiers and creates a child based upon it. Experiments indicate that the RC technique improves the performance of the learning system in fulfilling the given tasks. Moreover, it requires less learning time and a significantly smaller number of classifiers than the classical XCS with either binary or real value inputs.
 
|Beschreibung EN=Accuracy-based Learning Classifier System (XCS) is an online genetic-based machine learning method introduced by Wilson in 1995. It learns by maintaining several rules called classifiers in form of an input-output pair, where the input is a series of “genetic” information. XCS has the goals to learn the given task by responding to the environmental with a proper action and maximizing the awarded feedback. It also aims to generalize the rules in the population in order to reduce the usage of resources, along with an effort to minimize the time requirement.
 
|Beschreibung EN=Accuracy-based Learning Classifier System (XCS) is an online genetic-based machine learning method introduced by Wilson in 1995. It learns by maintaining several rules called classifiers in form of an input-output pair, where the input is a series of “genetic” information. XCS has the goals to learn the given task by responding to the environmental with a proper action and maximizing the awarded feedback. It also aims to generalize the rules in the population in order to reduce the usage of resources, along with an effort to minimize the time requirement.
  
XCS employs genetic operators (i.e., crossover and mutation) to discover general and accurate rules. However, these techniques have some drawbacks due to some randomizations (e.g., when picking crossover points) that potentially slow down the discovery process. A heuristic-based genetic operator called Rule Combining (RC) is introduced to address this issue. Instead of involving some random values, it generalizes rules by performing inductive reasoning on a pair of experienced classifiers and creates a child based upon it. Comparisons indicate that the RC technique is capable of improving the performance of XCS in fulfilling the given tasks. In particular, experiments indicate that it requires less learning time and a significantly smaller number of classifiers than the classical XCS.
+
XCS employs genetic operators (i.e., crossover and mutation) to discover general and accurate rules. However, these techniques have some drawbacks due to some randomizations (e.g., when picking crossover points) that potentially slow down the discovery process. A heuristic-based genetic operator called Rule Combining (RC) is introduced to address this issue. Instead of involving some random values, it generalizes rules by performing inductive reasoning on a pair of experienced classifiers and creates a child based upon it. Experiments indicate that the RC technique improves the performance of the learning system in fulfilling the given tasks. Moreover, it requires less learning time and a significantly smaller number of classifiers than the classical XCS with either binary or real value inputs.
 
|Veranstaltungsart=Graduiertenkolloquium
 
|Veranstaltungsart=Graduiertenkolloquium
 
|Start=2012/07/20 11:30:00
 
|Start=2012/07/20 11:30:00
 
|Ende=2012/07/20 12:30:00
 
|Ende=2012/07/20 12:30:00
 +
|Gebäude=11.40
 +
|Raum=214
 
|Vortragender=Nugroho Fredivianus
 
|Vortragender=Nugroho Fredivianus
 
|Eingeladen durch=Hartmut Schmeck
 
|Eingeladen durch=Hartmut Schmeck
 +
|PDF=Fredivianus Graduiertenkolloquium.pdf
 
|Forschungsgruppe=Effiziente Algorithmen
 
|Forschungsgruppe=Effiziente Algorithmen
 
|In News anzeigen=True
 
|In News anzeigen=True
 
}}
 
}}

Aktuelle Version vom 17. Juli 2012, 11:25 Uhr

A Heuristic-based Genetic Operator in Classifier Systems

Veranstaltungsart:
Graduiertenkolloquium




Accuracy-based Learning Classifier System (XCS) is an online genetic-based machine learning method introduced by Wilson in 1995. It learns by maintaining several rules called classifiers in form of an input-output pair, where the input is a series of “genetic” information. XCS has the goals to learn the given task by responding to the environmental with a proper action and maximizing the awarded feedback. It also aims to generalize the rules in the population in order to reduce the usage of resources, along with an effort to minimize the time requirement.

XCS employs genetic operators (i.e., crossover and mutation) to discover general and accurate rules. However, these techniques have some drawbacks due to some randomizations (e.g., when picking crossover points) that potentially slow down the discovery process. A heuristic-based genetic operator called Rule Combining (RC) is introduced to address this issue. Instead of involving some random values, it generalizes rules by performing inductive reasoning on a pair of experienced classifiers and creates a child based upon it. Experiments indicate that the RC technique improves the performance of the learning system in fulfilling the given tasks. Moreover, it requires less learning time and a significantly smaller number of classifiers than the classical XCS with either binary or real value inputs.

(Nugroho Fredivianus)




Start: 20. Juli 2012 um 11:30
Ende: 20. Juli 2012 um 12:30


Im Gebäude 11.40, Raum: 214

Veranstaltung vormerken: (iCal)


Veranstalter: Forschungsgruppe(n) Effiziente Algorithmen
Information: Media:Fredivianus Graduiertenkolloquium.pdf