Home > Research > Publications & Outputs > Temporal-difference Learning with Sampling Base...

Associated organisational unit

Electronic data

  • 2018-4

    Rights statement: Copyright c 2018, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.

    Accepted author manuscript, 1 MB, PDF-document

    Available under license: CC BY-NC: Creative Commons Attribution-NonCommercial 4.0 International License

Links

View graph of relations

Temporal-difference Learning with Sampling Baseline for Image Captioning

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSNConference contribution/Paper

Published
Close
Publication date1/02/2018
Host publication32nd AAAI Conference on Artificial Intelligence 2018
Place of PublicationPalo Alto
PublisherAAAI
Pages6706-6713
Number of pages8
ISBN (Print)9781577358008
Original languageEnglish

Abstract

The existing methods for image captioning usually train the language model under the cross entropy loss, which results in the exposure bias and inconsistency of evaluation metric. Recent research has shown these two issues can be well addressed by policy gradient method in reinforcement learning domain attributable to its unique capability of directly optimizing the discrete and non-differentiable evaluation metric. In this paper, we utilize reinforcement learning method to train the image captioning model. Specifically, we train our image captioning model to maximize the overall reward of the sentences by adopting the temporal-difference (TD) learning method, which takes the correlation between temporally successive actions into account. In this way, we assign different values to different words in one sampled sentence by a discounted coefficient when back-propagating the gradient with the REINFORCE algorithm, enabling the correlation between actions to be learned. Besides, instead of estimating a "baseline" to normalize the rewards with another network, we utilize the reward of another Monte-Carlo sample as the "baseline" to avoid high variance. We show that our proposed method can improve the quality of generated captions and outperforms the state-of-the-art methods on the benchmark dataset MS COCO in terms of seven evaluation metrics.

Bibliographic note

Copyright c 2018, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.