mlsanityData to 7 Oct 2026

Self-training Improves Pre-training for Few-shot Learning in Task-oriented Dialog Systems

arXiv 2108.12589 ↗First posted 28 Aug 2021cs.CLcited by 18

On the map: Synthetic data and self-improvement for LLMs → Self-training for few-shot NLP

Abstract (arXiv)

As the labeling cost for different modules in task-oriented dialog (ToD) systems is expensive, a major challenge is to train different modules with the least amount of labeled data. Recently, large-scale pre-trained language models, have shown promising results for few-shot learning in ToD. In this paper, we devise a self-training approach to utilize the abundant unlabeled dialog data to further improve state-of-the-art pre-trained models in few-shot learning scenarios for ToD systems. Specifically, we propose a self-training approach that iteratively labels the most confident unlabeled data to train a stronger Student model. Moreover, a new text augmentation technique (GradAug) is proposed to better train the Student by replacing non-crucial tokens using a masked language model. We conduct extensive experiments and present analyses on four downstream tasks in ToD, including intent classification, dialog state tracking, dialog act prediction, and response selection. Empirical results demonstrate that the proposed self-training approach consistently improves state-of-the-art pre-trained models (BERT, ToD-BERT) when only a small number of labeled data are available.

Key facts. Cited by 18 papers in the map (data through 7 Oct 2026). Placed in Self-training for few-shot NLP, one of 13 directions in Synthetic data and self-improvement for LLMs. Builds on 42 of its 66 references that are in the map. First posted on arXiv 28 Aug 2021.

Builds on

42 of 66 references.

References · 42

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
  4. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,266
  5. 2017Axiomatic Attribution for Deep Networkscited by 3,350
  6. 2013Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Mapscited by 2,348
  7. 2019MixMatch: A Holistic Approach to Semi-Supervised Learningcited by 1,390
  8. 2019Self-training with Noisy Student improves ImageNet classificationcited by 1,304
  9. 2018Hierarchical Neural Story Generationcited by 1,182
  10. 2017SmoothGrad: removing noise by adding noisecited by 1,097
  11. 2016Temporal Ensembling for Semi-Supervised Learningcited by 1,054
  12. 2017Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learningcited by 918
30 more
  1. 2019EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Taskscited by 800
  2. 2018MultiWOZ -- A Large-Scale Multi-Domain Wizard-of-Oz Dataset for Task-Oriented Dialogue Modellingcited by 760
  3. 2019Unified Language Model Pre-training for Natural Language Understanding and Generationcited by 668
  4. 2019An Evaluation Dataset for Intent Classification and Out-of-Scope Predictioncited by 378
  5. 2020Rethinking Pre-training and Self-trainingcited by 352
  6. 2020A Simple Language Model for Task-Oriented Dialoguecited by 331
  7. 2016Adversarial Training Methods for Semi-Supervised Text Classificationcited by 280
  8. 2019Billion-scale semi-supervised learning for image classificationcited by 232
  9. 2019Transferable Multi-Domain State Generator for Task-Oriented Dialogue Systemscited by 215
  10. 2020MixText: Linguistically-Informed Interpolation of Hidden Space for Semi-Supervised Text Classificationcited by 186
  11. 2020TOD-BERT: Pre-trained Natural Language Understanding for Task-Oriented Dialoguecited by 185
  12. 2019Revisiting Self-Training for Neural Sequence Generationcited by 154
  13. 2020Improved Noisy Student Training for Automatic Speech Recognitioncited by 139
  14. 2019Self-Training for End-to-End Speech Recognitioncited by 128
  15. 2019End-to-end ASR: from Supervised to Semi-Supervised Learning with Modern Architecturescited by 120
  16. 2019Hello, It's GPT-2 -- How Can I Help You? Towards the Use of Pretrained Language Models for Task-Oriented Dialogue Systemscited by 110
  17. 2020Self-training Improves Pre-training for Natural Language Understandingcited by 96
  18. 2020SSMBA: Self-Supervised Manifold Based Data Augmentation for Improving Out-of-Domain Robustnesscited by 94
  19. 2020Few-shot Natural Language Generation for Task-Oriented Dialogcited by 89
  20. 2019Learning to Self-Train for Semi-Supervised Few-Shot Classificationcited by 83
  21. 2020SOLOIST: Building Task Bots at Scale with Transfer Learning and Machine Teachingcited by 71
  22. 2021Leveraging Slot Descriptions for Zero-Shot Cross-Domain Dialogue State Trackingcited by 61
  23. 2020SlimIPL: Language-Model-Free Iterative Pseudo-Labelingcited by 54
  24. 2019Training Neural Response Selection for Task-Oriented Dialogue Systemscited by 48
  25. 2020Continual Learning for Natural Language Generation in Task-oriented Dialog Systemscited by 47
  26. 2020Uncertainty-aware Self-training for Text Classification with Few Labelscited by 19
  27. 2020A Self-Training Method for Machine Reading Comprehension with Soft Evidence Extractioncited by 9
  28. 2018Reinforced Co-Trainingcited by 7
  29. 2019Towards Universal Dialogue Act Tagging for Task-Oriented Dialoguescited by 7
  30. 2018Learning How to Self-Learn: Enhancing Self-Training Using Neural Reinforcement Learningcited by 2

Built on by

Cited by year

Citing papers per year: 2021: 4; 2022: 6; 2023: 4; 2024: 4.

18 papers.

Sort:

  1. 2022Improving In-Context Few-Shot Learning via Self-Supervised Trainingcited by 56
  2. 2022InstructDial: Improving Zero and Few-shot Generalization in Dialogue through Instruction Tuningcited by 39
  3. 2021CINS: Comprehensive Instruction for Few-shot Learning in Task-oriented Dialog Systemscited by 30
  4. 2021Unsupervised Neural Machine Translation with Generative Language Models Onlycited by 24
  5. 2022"Do you follow me?": A Survey of Recent Approaches in Dialogue State Trackingcited by 22
  6. 2021LMTurk: Few-Shot Learners as Crowdsourcing Workers in a Language-Model-as-a-Service Frameworkcited by 8
  7. 2021Self-Training with Differentiable Teachercited by 7
  8. 2023Prompter: Zero-shot Adaptive Prefixes for Dialogue State Tracking Domain Adaptationcited by 7
  9. 2023Stabilized In-Context Learning with Pre-trained Language Models for Few Shot Dialogue State Trackingcited by 6
  10. 2023Cache me if you Can: an Online Cost-aware Teacher-Student framework to Reduce the Calls to Large Language Modelscited by 4
  11. 2024Domain Adaptation in Intent Classification Systems: A Reviewcited by 4
  12. 2023SQATIN: Supervised Instruction Tuning Meets Question Answering for Improved Dialogue NLUcited by 3
6 more
  1. 2022Self-training with Two-phase Self-augmentation for Few-shot Dialogue Generationcited by 1
  2. 2022CSS: Combining Self-training and Self-supervised Learning for Few-shot Dialogue State Trackingcited by 1
  3. 2022Self-Training with Purpose Preserving Augmentation Improves Few-shot Generative Dialogue State Trackingcited by 0
  4. 2024State Value Generation with Prompt Learning and Self-Training for Low-Resource Dialogue State Trackingcited by 0
  5. 2024Comparing Data Augmentation Methods for End-to-End Task-Oriented Dialog Systemscited by 0
  6. 2024A Comparative Study of Pre-training and Self-trainingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2021Revisiting Self-Training for Few-Shot Learning of Language Model
  2. 2024SynthDST: Synthetic Data is All You Need for Few-Shot Dialog State Tracking
  3. 2021STraTA: Self-Training with Task Augmentation for Better Few-shot Learning
  4. 2024Benchmarks Underestimate the Readiness of Multi-lingual Dialogue Agents
  5. 2024Self-seeding and Multi-intent Self-instructing LLMs for Generating Intent-aware Information-Seeking dialogs
  6. 2024Towards Zero-Shot, Controllable Dialog Planning with LLMs
  7. 2023Knowledge-Retrieval Task-Oriented Dialog Systems with Semi-Supervision
  8. 2024Self-Supervised Alignment with Mutual Information: Learning to Follow Principles without Preference Labels