A two step algorithm for learning from unspecific reinforcement

• Verfasst von: Kühn, Reimer [VerfasserIn]
• Verfasst von: Stamatescu, Ion-Olimpiu [VerfasserIn]
• Titel: A two step algorithm for learning from unspecific reinforcement
• Verf.angabe: Reimer Kühn, Ion-Olimpiu Stamatescu
• Fussnoten: Gesehen am 06.03.2018
• Titel Quelle: Enthalten in: De.arxiv.org
• Jahr Quelle: 1999
• Band/Heft Quelle: (1999) Artikel-Nummer 9902354, 13 Seiten
• Datenträger: Online-Ressource
• Sprache: eng
• K10plus-PPN: 157044059X

Abstract

We study a simple learning model based on the Hebb rule to cope with "delayed", unspecific reinforcement. In spite of the unspecific nature of the information-feedback, convergence to asymptotically perfect generalization is observed, with a rate depending, however, in a non- universal way on learning parameters. Asymptotic convergence can be as fast as that of Hebbian learning, but may be slower. Moreover, for a certain range of parameter settings, it depends on initial conditions whether the system can reach the regime of asymptotically perfect generalization, or rather approaches a stationary state of poor generalization.