Dialogue Transformers
By Vladimir Vlasov, Johannes E. M. Mosig, Alan Nichol. 2019
63 citations recorded by the source corpus.
We introduce a dialogue policy based on a transformer architecture, where the self-attention mechanism operates over the sequence of dialogue turns. Recent work has used hierarchical recurrent neural networks to encode multiple utterances in a dialogue context, but we argue that a pure self-attention mechanism is more suitable. By default, an RNN assumes that every item in a sequence is relevant for producing an encoding of the full sequence, but a single conversation can consist of multiple overlapping discourse segments as speakers interleave multiple topics. A transformer picks which turns to include in its encoding of the current dialogue state, and is naturally suited to selectively ignoring or attending to dialogue history. We compare the performance of the Transformer Embedding Dialogue (TED) policy to an LSTM and to the REDP, which was specifically designed to overcome this limitation of RNNs.
Subfields: Artificial Intelligence
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Ivan Vulic
- Iñigo Casanueva
- Pei-hao Su
- Michael Heck
- Carel van Niekerk
- Nurul Lubis
- Milica Gavsi'c
- Christian Geishauser
Citations in the corpus
- Dynamic Dialogue Policy for Continual Reinforcement Learning (2022, background): “Works that allow new actions to be added to the action set compare the encoded state and action embeddings with each other (Lee, 2017; Xu et al., 2020; Vlasov et al., 2019), suggesting that exploiting similarities is key not only for state representations but also for action prediction.”
- ConveRT: Efficient and Accurate Conversational Representations from Transformers (2019, background): “Note that more sophisticated solutions to fusing dialog history are possible such as using attention over older contexts as done by Vlasov et al. (2019) on the much smaller MultiWOZ 2.1 dataset (Eric et al., 2019), but we have opted for simple concatenation as an efficient solution for training on…”
- Dynamic Dialogue Policy Transformer for Continual Reinforcement Learning (2022, background): “Works that allow new actions to be added to the action set compare the encoded state and action embeddings with each other (Lee, 2017; Xu et al., 2020; Vlasov et al., 2019), suggesting that exploiting similarities is key not only for state representations but also for action prediction.”