CoChat

Learning from Dialogue after Deployment: Feed Yourself, Chatbot!

Open access
Hancock, Braden, Bordes, Antoine, Mazare, Pierre-Emmanuel, Weston, Jason · 2019 · Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics

The DOI matches its Crossref record; 55 of the 55 references it cites were found in a registry or an index.

Abstract

The majority of conversations a dialogue agent sees over its lifetime occur after it has already been trained and deployed, leaving a vast store of potential training signal untapped.In this work, we propose the self-feeding chatbot, a dialogue agent with the ability to extract new training examples from the conversations it participates in.As our agent engages in conversation, it also estimates user satisfaction in its responses.When the conversation appears to be going well, the user's responses become new training examples to imitate.When the agent believes it has made a mistake, it asks for feedback; learning to predict the feedback that will be given improves the chatbot's dialogue abilities further.On the PERSONACHAT chitchat dataset with over 131k training examples, we find that learning from dialogue with a selffeeding chatbot significantly improves performance, regardless of the amount of traditional supervision.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.18653/v1/p19-1358

Cites

55 of 55 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 As the self-feeding chatbot engages in dialogue, it estimates user satisfaction to know when to ask for feedback. From the satisfied responses and feedback responses, new training examples are extracted for the DIALOGUE and FEEDBACK tasks, respectively, both of which improve the model's dialogue abilities further.
Figure 1 As the self-feeding chatbot engages in dialogue, it estimates user satisfaction to know when to ask for feedback. From the satisfied responses and feedback responses, new training examples are extracted for the DIALOGUE and FEEDBACK tasks, respectively, both of which improve the model's dialogue abilities further.CC-BY · From the PDF
Figure 2 (1) The chatbot is first trained with any available supervised data (boxed in red) on the Human-Human (HH) DIALOGUE ( x, y ) HH and SATISFACTION ( x, s ) tasks. (2) During deployment, whenever the predicted satisfaction score of the current conversation x is above the threshold ( ˆ s > t ), a new Human-Bot (HB) DIALOGUE example ( x, y ) HB is extracted and the bot continues the conversation with its own response ˆ y . Otherwise, the chatbot requests feedback with question q and extracts a new FEEDBACK example ( x, f ) . (3) The chatbot is periodically retrained with the available examples from all four datasets, improving its DIALOGUE performance without collecting any new supervised examples.
Figure 2 (1) The chatbot is first trained with any available supervised data (boxed in red) on the Human-Human (HH) DIALOGUE ( x, y ) HH and SATISFACTION ( x, s ) tasks. (2) During deployment, whenever the predicted satisfaction score of the current conversation x is above the threshold ( ˆ s > t ), a new Human-Bot (HB) DIALOGUE example ( x, y ) HB is extracted and the bot continues the conversation with its own response ˆ y . Otherwise, the chatbot requests feedback with question q and extracts a new FEEDBACK example ( x, f ) . (3) The chatbot is periodically retrained with the available examples from all four datasets, improving its DIALOGUE performance without collecting any new supervised examples.CC-BY · From the PDF
Figure 3 The first 20k examples for all models are supervised DIALOGUE examples. This model is deployed to collect 20k FEEDBACK examples (set A). If the model is retrained before collecting the next 20k examples (set B), the fresher feedback results in better performance ( p = 0 . 027) . Shaded regions depict 95% confidence intervals.
Figure 3 The first 20k examples for all models are supervised DIALOGUE examples. This model is deployed to collect 20k FEEDBACK examples (set A). If the model is retrained before collecting the next 20k examples (set B), the fresher feedback results in better performance ( p = 0 . 027) . Shaded regions depict 95% confidence intervals.CC-BY · From the PDF
Image 4
CC-BY · From the PDF
Image 5
CC-BY · From the PDF

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Research works printed but not linked: 3

  1. T. Hong, O. Kwon, and Y. Kim. 2019. An end-to- end trainable task-oriented dialog system with hu- man feedback. In Association for the Advancement of Artificial Intelligence (AAAI).
  2. J. Li, A. H. Miller, S. Chopra, M. Ranzato, and J. We- ston. 2017a. Dialogue learning with human-in-the- loop. In International Conference on Learning Rep- resentations (ICLR).
  3. S. Tong and D. Koller. 2001. Support vector ma- chine active learning with applications to text clas- sification. Journal of machine learning research, 2(0):45-66.

Cited by

See all 168

Metadata

Authors
Hancock, B., Bordes, A., Mazare, P.-E., & Weston, J.
Affiliations
  • Stanford University (US)
  • Meta (United Kingdom) (GB)
Venue
Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics
Published
2019
Pages
3667-3684
ISSN
0736-587X (ISSN-L)
OpenAlex
W2910458567
Cited by
161
References
53
Citations by year
2026: 6 · 2025: 14 · 2024: 13 · 2023: 21 · 2022: 23 · 2021: 41 · 2020: 33 · 2019: 10
FWCI
10.24
Type
conference paper
Language
en
Open access
Licence
cc-by
Locations
  • Proceedings of the conference - Association for Computational Linguistics. Meeting (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
  • Physical Sciences › Computer Science › Artificial Intelligence › Intelligent Tutoring Systems and Adaptive Learning
  • Social Sciences › Business, Management and Accounting › Marketing › AI in Service Interactions
Keywords
imitation learning

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.