Improved Image Captioning via Policy Gradient optimization of SPIDEr

TL;DR

This paper introduces a policy gradient method to directly optimize a combined semantic and syntactic image captioning metric, SPIDEr, resulting in captions that better align with human preferences and improve over previous training methods.

Contribution

It proposes a novel policy gradient approach using Monte Carlo rollouts to optimize the SPIDEr metric for image captioning, outperforming prior methods like MIXER.

Findings

Optimized captions are more semantically faithful and syntactically fluent.

The method yields captions preferred by human raters over traditional MLE-trained models.

The approach simplifies optimization and improves results across metrics.

Abstract

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 human assessments of quality. Standard syntactic evaluation metrics, such as BLEU, METEOR and ROUGE, are also not well correlated. The newer SPICE and CIDEr metrics are better correlated, but have traditionally been hard to optimize for. In this paper, we show how to use a policy gradient (PG) method to directly optimize a linear combination of SPICE and CIDEr (a combination we call SPIDEr): the SPICE score ensures our captions are semantically faithful to the image, while CIDEr score ensures our captions are syntactically fluent. The PG method we propose improves on the prior MIXER approach, by using Monte Carlo rollouts instead of mixing MLE training with PG. We show empirically that our algorithm leads to easier optimization and improved results compared to MIXER. Finally, we show that using our PG method we can optimize any of the metrics, including the proposed SPIDEr metric which results in image captions that are strongly preferred by human raters compared to captions generated by the same model but trained to optimize MLE or the COCO metrics.