Title: Learning long-term dependencies in segmented-memory recurrent neural networks with backpropagation of error
Authors : Glüge, Stefan
Böck, Ronald
Palm, Günther
Wendemuth, Andreas
Published in : Neurocomputing
Volume(Issue) : 141
Pages : 54
Pages to: 64
Publisher / Ed. Institution : Elsevier
Issue Date: 2014
License (according to publishing contract) : Licence according to publishing contract
Type of review: Peer review (publication)
Language : English
Subject (DDC) : 004: Computer science
Abstract: In general, recurrent neural networks have difficulties in learning long-term dependencies. The segmented-memory recurrent neural network (SMRNN) architecture together with the extended real-time recurrent learning (eRTRL) algorithm was proposed to circumvent this problem. Due to its computational complexity eRTRL becomes impractical with increasing network size. Therefore, we introduce the less complex extended backpropagation through time (eBPTT) for SMRNN together with a layer-local unsupervised pre-training procedure. A comparison on the information latching problem showed that eRTRL is better able to handle the latching of information over longer periods of time, even though eBPTT guaranteed a better generalisation when training was successful. Further, pre-training significantly improved the ability to learn long-term dependencies with eBPTT. Therefore, the proposed eBPTT algorithm is suited for tasks that require big networks where eRTRL is impractical. The pre-training procedure itself is independent of the supervised learning algorithm and can improve learning in SMRNN in general.
Departement: Life Sciences and Facility Management
Organisational Unit: Institute of Applied Simulation (IAS)
Publication type: Article in scientific journal
DOI : 10.1016/j.neucom.2013.11.043
ISSN: 0925-2312
1872-8286
URI: https://digitalcollection.zhaw.ch/handle/11475/15300
Appears in Collections:Publikationen Life Sciences und Facility Management

Files in This Item:
There are no files associated with this item.


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.