In this book, we propose novel deterministic RNN training algorithms that adopt a nonmonotone approach. This allows learning behaviour to deteriorate in some iterations; nevertheless the overall learning performance is improved over time. The nonmonotone RNN training methods, which take their theoretical basis from the theory of deterministic nonlinear optimisation, aim at better exploring the search space and enhancing the convergence behaviour of gradient-based methods. They generate nonmonotone behaviour by incorporating conditions that employ forcing functions, which are used to measure the sufficiency of error reduction, and an adaptive window, whose size is informed by estimating the morphology of the error surface locally. The thesis develops nonmonotone 1st- and 2nd-order methods and discusses their convergence properties. The proposed algorithms are applied to training RNNs of various sizes and architectures, namely Feed-Forward Time-Delay networks, Elman Networks and Nonlinear Autoregressive Networks with Exogenous Inputs Networks, in symbolic sequence processing problems. Numerical results show that the proposed nonmonotone learning algorithms train more effectively.
Die Inhaltsangabe kann sich auf eine andere Ausgabe dieses Titels beziehen.
After granted his PhD (Neural Networks Learning) from the Computer Science and Information Systems Department of the Birkbeck College, University of London, England in 2011, Chun-Cheng Peng is currently a Post-doctoral Research Associate in National Chin-Yi University of Technology, Taiwan. His research interests are mainly in nonmonotone learning.
„Über diesen Titel“ kann sich auf eine andere Ausgabe dieses Titels beziehen.
EUR 28,83 für den Versand von Vereinigtes Königreich nach Deutschland
Versandziele, Kosten & DauerGratis für den Versand innerhalb von/der Deutschland
Versandziele, Kosten & DauerAnbieter: moluna, Greven, Deutschland
Zustand: New. Dieser Artikel ist ein Print on Demand Artikel und wird nach Ihrer Bestellung fuer Sie gedruckt. Autor/Autorin: Peng Chun-ChengAfter granted his PhD (Neural Networks Learning) from the Computer Science and Information Systems Department of the Birkbeck College, University of London, England in 2011, Chun-Cheng Peng is currently a Post-doctoral. Bestandsnummer des Verkäufers 5502357
Anzahl: Mehr als 20 verfügbar
Anbieter: buchversandmimpf2000, Emtmannsberg, BAYE, Deutschland
Taschenbuch. Zustand: Neu. Neuware -In this book, we propose novel deterministic RNN training algorithms that adopt a nonmonotone approach. This allows learning behaviour to deteriorate in some iterations; nevertheless the overall learning performance is improved over time. The nonmonotone RNN training methods, which take their theoretical basis from the theory of deterministic nonlinear optimisation, aim at better exploring the search space and enhancing the convergence behaviour of gradient-based methods. They generate nonmonotone behaviour by incorporating conditions that employ forcing functions, which are used to measure the sufficiency of error reduction, and an adaptive window, whose size is informed by estimating the morphology of the error surface locally. The thesis develops nonmonotone 1st- and 2nd-order methods and discusses their convergence properties. The proposed algorithms are applied to training RNNs of various sizes and architectures, namely Feed-Forward Time-Delay networks, Elman Networks and Nonlinear Autoregressive Networks with Exogenous Inputs Networks, in symbolic sequence processing problems. Numerical results show that the proposed nonmonotone learning algorithms train more effectively.Books on Demand GmbH, Überseering 33, 22297 Hamburg 256 pp. Englisch. Bestandsnummer des Verkäufers 9783846599532
Anzahl: 2 verfügbar
Anbieter: AHA-BUCH GmbH, Einbeck, Deutschland
Taschenbuch. Zustand: Neu. nach der Bestellung gedruckt Neuware - Printed after ordering - In this book, we propose novel deterministic RNN training algorithms that adopt a nonmonotone approach. This allows learning behaviour to deteriorate in some iterations; nevertheless the overall learning performance is improved over time. The nonmonotone RNN training methods, which take their theoretical basis from the theory of deterministic nonlinear optimisation, aim at better exploring the search space and enhancing the convergence behaviour of gradient-based methods. They generate nonmonotone behaviour by incorporating conditions that employ forcing functions, which are used to measure the sufficiency of error reduction, and an adaptive window, whose size is informed by estimating the morphology of the error surface locally. The thesis develops nonmonotone 1st- and 2nd-order methods and discusses their convergence properties. The proposed algorithms are applied to training RNNs of various sizes and architectures, namely Feed-Forward Time-Delay networks, Elman Networks and Nonlinear Autoregressive Networks with Exogenous Inputs Networks, in symbolic sequence processing problems. Numerical results show that the proposed nonmonotone learning algorithms train more effectively. Bestandsnummer des Verkäufers 9783846599532
Anzahl: 1 verfügbar
Anbieter: BuchWeltWeit Ludwig Meier e.K., Bergisch Gladbach, Deutschland
Taschenbuch. Zustand: Neu. This item is printed on demand - it takes 3-4 days longer - Neuware -In this book, we propose novel deterministic RNN training algorithms that adopt a nonmonotone approach. This allows learning behaviour to deteriorate in some iterations; nevertheless the overall learning performance is improved over time. The nonmonotone RNN training methods, which take their theoretical basis from the theory of deterministic nonlinear optimisation, aim at better exploring the search space and enhancing the convergence behaviour of gradient-based methods. They generate nonmonotone behaviour by incorporating conditions that employ forcing functions, which are used to measure the sufficiency of error reduction, and an adaptive window, whose size is informed by estimating the morphology of the error surface locally. The thesis develops nonmonotone 1st- and 2nd-order methods and discusses their convergence properties. The proposed algorithms are applied to training RNNs of various sizes and architectures, namely Feed-Forward Time-Delay networks, Elman Networks and Nonlinear Autoregressive Networks with Exogenous Inputs Networks, in symbolic sequence processing problems. Numerical results show that the proposed nonmonotone learning algorithms train more effectively. 256 pp. Englisch. Bestandsnummer des Verkäufers 9783846599532
Anzahl: 2 verfügbar
Anbieter: Books Puddle, New York, NY, USA
Zustand: New. pp. 256. Bestandsnummer des Verkäufers 2698162546
Anzahl: 4 verfügbar
Anbieter: Biblios, Frankfurt am main, HESSE, Deutschland
Zustand: New. PRINT ON DEMAND pp. 256. Bestandsnummer des Verkäufers 1898162552
Anzahl: 4 verfügbar
Anbieter: Majestic Books, Hounslow, Vereinigtes Königreich
Zustand: New. Print on Demand pp. 256 2:B&W 6 x 9 in or 229 x 152 mm Perfect Bound on Creme w/Gloss Lam. Bestandsnummer des Verkäufers 95316141
Anzahl: 4 verfügbar
Anbieter: Mispah books, Redhill, SURRE, Vereinigtes Königreich
Paperback. Zustand: Like New. Like New. book. Bestandsnummer des Verkäufers ERICA75838465995306
Anzahl: 1 verfügbar