End-to-end Image Captioning Exploits Multimodal Distributional Similarity.
File(s)
Accepted version
Author(s)
Madhyastha, PS
Wang, J
Specia, L
Type
Conference Paper
Abstract
We hypothesize that end-to-end neural image captioning systems work seemingly well because they exploit and learn `distributional similarity' in a multimodal feature space by mapping a test image to similar training images in this space and generating a caption from the same space. To validate our hypothesis, we focus on the `image' side of image captioning, and vary the input image representation but keep the RNN text generation component of a CNN-RNN model constant. Our analysis indicates that image captioning models (i) are capable of separating structure from noisy input representations; (ii) suffer virtually no significant performance loss when a high dimensional representation is compressed to a lower dimensional space; (iii) cluster images with similar visual and linguistic information together. Our findings indicate that our distributional similarity hypothesis holds. We conclude that regardless of the image representation used image captioning systems seem to match images and generate captions in a learned joint image-text semantic subspace.
Date Issued
2018
Date Acceptance
2018-07-06
Citation
CoRR, 2018, abs/1809.04144
Publisher
BMVC
Journal / Book Title
CoRR
Volume
abs/1809.04144
Copyright Statement
© 2018. The copyright of this document resides with its authors.
It may be distributed unchanged freely in print or electronic forms.
It may be distributed unchanged freely in print or electronic forms.
Identifier
http://arxiv.org/abs/1809.04144
Source
29th British Machine Vision Conference
Subjects
cs.CV
Notes
Published in BMVC 2018
Publication Status
Published
Date Publish Online
2018-09-06