Learning with recurrent neural networks
Folding networks, a generalisation of recurrent neural networks to tree structured inputs, are investigated as a mechanism to learn regularities on classical symbolic data, for example. The architecture, the training mechanism, and several applications in different areas are explained. Afterwards a...
Na minha lista:
| Autor principal: | |
|---|---|
| Formato: | Livre numérique |
| Idioma: | Anglais |
| Publicado em: |
London ; Paris [etc.] :
Springer-Verlag London Limited : Springer e-books
[20..].
Cham : Springer Nature |
| Colecção: | Lecture notes in control and information sciences
254 |
| Assuntos: | |
| Acesso em linha: | Accès sur la plateforme de l'éditeur Accès sur la plateforme Istex Accès Université d'Orléans Accès INSA CVL |
| Nota: |
Archives Springer e-books (Licence nationale) Archives Springer e-books (Licence nationale) |
| Autres localisations: | Voir dans le Sudoc |
| Edition sous un autre format: | • Learning with recurrent neural networks, Barbara Hammer, London, Springer, 2000, X-148 p., Lecture notes in control and information sciences, 1-85233-343-X • Learning with Recurrent Neural Networks, Texte imprimé, 9781447139591 |
| Resumo: | Folding networks, a generalisation of recurrent neural networks to tree structured inputs, are investigated as a mechanism to learn regularities on classical symbolic data, for example. The architecture, the training mechanism, and several applications in different areas are explained. Afterwards a theoretical foundation, proving that the approach is appropriate as a learning mechanism in principle, is presented: Their universal approximation ability is investigated- including several new results for standard recurrent neural networks such as explicit bounds on the required number of neurons and the super Turing capability of sigmoidal recurrent networks. The information theoretical learnability is examined - including several contribution to distribution dependent learnability, an answer to an open question posed by Vidyasagar, and a generalisation of the recent luckiness framework to function classes. Finally, the complexity of training is considered - including new results on the loading problem for standard feedforward networks with an arbitrary multilayered architecture, a correlated number of neurons and training set size, a varying number of hidden neurons but fixed input dimension, or the sigmoidal activation function, respectively. |
|---|---|
| Descrição do item: | Archives Springer e-books (Licence nationale) Archives Springer e-books (Licence nationale) |
| ISBN: | 9781846285677 (PDF) 9781846285677 |
| ISSN: | 1610-7411 |
| Acesso: | Accès en ligne pour les établissements français bénéficiaires des licences nationales Accès soumis à abonnement pour tout autre établissement Conditions particulières de réutilisation pour les bénéficiaires des licences nationales. https://www.licencesnationales.fr/springer-nature-ebooks-contrat-licence-ln-2017 |