Multi-domain learning and generalization in dialog state tracking

By Jason Williams. 2013

59 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Milica Gasic
2. Tsung-Hsien Wen
3. Van-Khanh Tran
4. Pei-hao Su
5. Nikola Mrksic
6. Bing Liu
7. Ondrej Dusek
8. Matthew Henderson

Citations in the corpus

• Multi-domain Dialog State Tracking using Recurrent Neural Networks (2015, background): “, 2013), maximum entropy models (Williams, 2013) and web-style ranking (Williams, 2014).”
• Word-Based Dialog State Tracking with Recurrent Neural Networks (2014, methodology): “Whereas, most previous approaches to discriminative state tracking have adapted stationary classifiers to the temporal process of dialog (Bohus and Rudnicky, 2006; Lee and Eskenazi, 2013; Lee, 2013; Williams, 2013; Henderson et al., 2013b).”
• A Survey on Dialogue Systems: Recent Advances and New Frontiers (2017, background): “bability distribution over each slot for each turn. A variety of statistical approaches, including robust sets of handcrafted rules [93], conditional random fields [36; 35; 63], maximum entropy models [98] and web-style ranking [101] have emerged in Dialog State Tracking Challenge (DSTC) shared tasks”
• Incremental on-line adaptation of POMDP-based dialogue managers to extended domains (2014, background): “This adaptation must be robust to mismatched training data [7] and support adequate user modelling [8].”
• Natural Language Generation for Spoken Dialogue System using RNN Encoder-Decoder Networks (2017, methodology): “…tackle this problem, such as (Mrkˇsi´c et al., 2015) using RNN-based networks for multi-domain dialogue state tracking, (Wen et al., 2016a) using a procedure to train multi-domain via multiple adaptation steps, or (Ga ˇ si ´ c et al., 2015; Williams, 2013) adapting of SDS components to new domains.”
• Machine Learning for Dialog State Tracking: A Review (2015, background): “If a small amount of training exists, then multi-domain learning can be exploited when adapting to the new domain [32].”
• Multi-task Learning for Joint Language Understanding and Dialogue State Tracking (2018, background): “, 2014), other maximum entropy models (Williams, 2013) and web-style ranking (Williams, 2014).”
• Recent advances in deep learning based dialogue systems: a systematic survey (2021, methodology): “Statistical methods (Lee, 2013; Lee and Eskenazi, 2013; Ren et al., 2013; Williams, 2013, 2014) also suffer from noisy conditions and ambiguity (Young et al., 2010).”
• The Dialog State Tracking Challenge Series: A Review (2016, background): “V2.CA measures (Williams et al., 2013).”