Generative Conversational Networks

Inspired by recent work in meta-learning and generative teaching networks, we propose a framework called Generative Conversational Networks, in which conversational agents learn to generate their own labelled training data (given some seed data) and then train themselves from that data to perform a given task. We use reinforcement learning to optimize the data generation process where the reward signal is the agent’s performance on the task. The task can be any language-related task, from intent detection to full task-oriented conversations. In this work, we show that our approach is able to generalise from seed data and performs well in limited data and limited computation settings, with significant gains for intent detection and slot tagging across multiple datasets: ATIS, TOD, SNIPS, and Restaurants8k. We show an average improvement of 35% in intent detection and 21% in slot tagging over a baseline model trained from the seed data. We also conduct an analysis of the novelty of the generated data and provide generated examples for intent detection, slot tagging, and non-goal oriented conversations.

[1]  Dilek Z. Hakkani-Tür,et al.  Topical-Chat: Towards Knowledge-Grounded Open-Domain Conversations , 2019, INTERSPEECH.

[2]  Pascale Fung,et al.  Coach: A Coarse-to-Fine Approach for Cross-domain Slot Filling , 2020, ACL.

[3]  Romain Laroche,et al.  Reward Shaping for Statistical Optimisation of Dialogue Management , 2013, SLSP.

[4]  George R. Doddington,et al.  The ATIS Spoken Language Systems Pilot Corpus , 1990, HLT.

[5]  Monica S. Lam,et al.  Zero-Shot Transfer Learning with Synthesized Data for Multi-Domain Dialogue State Tracking , 2020, ACL.

[6]  Jianfeng Gao,et al.  Few-shot Natural Language Generation for Task-Oriented Dialog , 2020, FINDINGS.

[7]  Kenton Lee,et al.  Neural Data Augmentation via Example Extrapolation , 2021, ArXiv.

[8]  Zhihan Zhou,et al.  Few-shot Slot Tagging with Collapsed Dependency Transfer and Label-enhanced Task-adaptive Projection Network , 2020, ACL.

[9]  Alec Radford,et al.  Fine-Tuning Language Models from Human Preferences , 2019, ArXiv.

[10]  Zhiyu Chen,et al.  Few-shot NLG with Pre-trained Language Model , 2020, ACL.

[11]  Jianfeng Gao,et al.  Data Augmentation for Spoken Language Understanding via Pretrained Models , 2020, ArXiv.

[12]  Chien-Wei Lin,et al.  Few Shot Dialogue State Tracking using Meta-learning , 2021, ArXiv.

[13]  Richard Socher,et al.  Transferable Multi-Domain State Generator for Task-Oriented Dialogue Systems , 2019, ACL.

[14]  Shashi Narayan,et al.  Leveraging Pre-trained Checkpoints for Sequence Generation Tasks , 2019, Transactions of the Association for Computational Linguistics.

[15]  Le-Minh Nguyen,et al.  Adversarial Domain Adaptation for Variational Neural Language Generation in Dialogue Systems , 2018, COLING.

[16]  Francesco Caltagirone,et al.  Snips Voice Platform: an embedded Spoken Language Understanding system for private-by-design voice interfaces , 2018, ArXiv.

[17]  Omer Levy,et al.  RoBERTa: A Robustly Optimized BERT Pretraining Approach , 2019, ArXiv.

[18]  Aman Alok,et al.  Protoda: Efficient Transfer Learning for Few-Shot Intent Classification , 2021, 2021 IEEE Spoken Language Technology Workshop (SLT).

[19]  Boi Faltings,et al.  Meta-Learning for Low-resource Natural Language Generation in Task-oriented Dialogue Systems , 2019, IJCAI.

[20]  Joel Lehman,et al.  Generative Teaching Networks: Accelerating Neural Architecture Search by Learning to Generate Synthetic Training Data , 2019, ICML.

[21]  David Vandyke,et al.  Reward Shaping with Recurrent Neural Networks for Speeding up On-Line Policy Learning in Spoken Dialogue Systems , 2015, SIGDIAL Conference.

[22]  Wanxiang Che,et al.  C2C-GenDA: Cluster-to-Cluster Generation for Data Augmentation of Slot Filling , 2020, AAAI.

[23]  Yann LeCun,et al.  The mnist database of handwritten digits , 2005 .

[24]  Amos Storkey,et al.  Meta-Learning in Neural Networks: A Survey , 2020, IEEE transactions on pattern analysis and machine intelligence.

[25]  Lei Zheng,et al.  Texygen: A Benchmarking Platform for Text Generation Models , 2018, SIGIR.

[26]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[27]  Matthew Henderson,et al.  Span-ConveRT: Few-shot Span Extraction for Dialog with Pretrained Conversational Representations , 2020, ACL.

[28]  Dilek Z. Hakkani-Tür,et al.  Robust Zero-Shot Cross-Domain Slot Filling with Example Values , 2019, ACL.

[29]  Alex Krizhevsky,et al.  Learning Multiple Layers of Features from Tiny Images , 2009 .

[30]  Sebastian Schuster,et al.  Cross-lingual Transfer Learning for Multilingual Task Oriented Dialog , 2018, NAACL.

[31]  Mihir Kale,et al.  Few-Shot Natural Language Generation by Rewriting Templates , 2020, ArXiv.