Please use this identifier to cite or link to this item:
Volltext verfügbar? / Dokumentlieferung
doi:10.22028/D291-30974
Title: | Improving Variational Encoder-Decoders in Dialogue Generation |
Author(s): | Shen, Xiaoyu Su, Hui Niu, Shuzi Demberg, Vera |
Language: | English |
Title: | Proceedings of the 2018 AAAIACM Conference on AI, Ethics, and Society |
Pages: | 9 |
Publisher/Platform: | ACM |
Year of Publication: | 2018 |
Place of publication: | New York, NY |
Title of the Conference: | AIES 2018 |
Place of the conference: | New Orleans, Louisiana, USA |
Publikation type: | Conference Paper |
Abstract: | Variational encoder-decoders (VEDs) have shown promising results in dialogue generation. However, the latent variable distributions are usually approximated by a much simpler model than the powerful RNN structure used for encoding and decoding, yielding the KL-vanishing problem and inconsistent training objective. In this paper, we separate the training step into two phases: The first phase learns to autoencode discrete texts into continuous embeddings, from which the second phase learns to generalize latent representations by reconstructing the encoded embedding. In this case, latent variables are sampled by transforming Gaussian noise through multi-layer perceptrons and are trained with a separate VED model, which has the potential of realizing a much more flexible distribution. We compare our model with current popular models and the experiment demonstrates substantial improvement in both metric-based and human evaluations. |
Link to this record: | hdl:20.500.11880/29740 http://dx.doi.org/10.22028/D291-30974 |
ISBN: | 978-1-4503-6012-8 |
Date of registration: | 24-Sep-2020 |
Faculty: | MI - Fakultät für Mathematik und Informatik |
Department: | MI - Informatik |
Professorship: | MI - Prof. Dr. Vera Demberg |
Collections: | SciDok - Der Wissenschaftsserver der Universität des Saarlandes |
Files for this record:
There are no files associated with this item.
Items in SciDok are protected by copyright, with all rights reserved, unless otherwise indicated.