Improved image captioning via policy gradient optimization of spider

S Liu, Z Zhu, N Ye, S Guadarrama… - Proceedings of the …, 2017 - openaccess.thecvf.com
Current image captioning methods are usually trained via maximum likelihood estimation.
However, the log-likelihood score of a caption does not correlate well with human …
被引用次数：503 相关文章

Improved Image Captioning via Policy Gradient optimization of SPIDEr

S Liu, Z Zhu, N Ye, S Guadarrama, K Murphy - arXiv preprint arXiv …, 2016 - arxiv.org
Current image captioning methods are usually trained via (penalized) maximum likelihood
estimation. However, the log-likelihood score of a caption does not correlate well with …

Improved Image Captioning via Policy Gradient optimization of SPIDEr

S Liu, Z Zhu, N Ye, S Guadarrama… - 2017 IEEE International …, 2017 - computer.org
Current image captioning methods are usually trained via maximum likelihood estimation.
However, the log-likelihood score of a caption does not correlate well with human …

Improved Image Captioning via Policy Gradient optimization of SPIDEr

S Liu, Z Zhu, N Ye, S Guadarrama… - 2017 IEEE International …, 2017 - ieeexplore.ieee.org
Current image captioning methods are usually trained via maximum likelihood estimation.
However, the log-likelihood score of a caption does not correlate well with human …
[PDF] researchgate.net

[PDF][PDF] Improved Image Captioning via Policy Gradient optimization of SPIDEr

S Liu, Z Zhu, N Ye, S Guadarrama, K Murphy - researchgate.net
Current image captioning methods are usually trained via (penalized) maximum likelihood
estimation. However, the log-likelihood score of a caption does not correlate well with …

Improved Image Captioning via Policy Gradient optimization of SPIDEr

S Liu, Z Zhu, N Ye, S Guadarrama, K Murphy - arXiv e-prints, 2016 - ui.adsabs.harvard.edu
Current image captioning methods are usually trained via (penalized) maximum likelihood
estimation. However, the log-likelihood score of a caption does not correlate well with …