Artificial Intelligence


Recurrent Capsule Network for Image Generation

Authors: Srikumar Sastry

We have already seen state-of-the-art image generation techniques with Generative Adversarial Networks (Goodfellow et al. 2014), Variational Autoencoder and Recurrent Network for Image generation (K. Gregor et al. 2015). But all these architectures fail to learn object location and pose in images. In this paper, I propose Recurrent Capsule Network based on variational auto encoding framework which can not only preserve equivariance in images in the latent space but also can be used for image classification and generation. For image classification, it can recognise highly overlapping objects due to the use of capsules (Hinton et al. 2011), considerably better than convolutional networks. It can generate images which can be difficult to differentiate from the real data.

Comments: 9 Pages.

Download: PDF

Submission history

[v1] 2018-04-07 11:12:53

Unique-IP document downloads: 813 times is a pre-print repository rather than a journal. Articles hosted may not yet have been verified by peer-review and should be treated as preliminary. In particular, anything that appears to include financial or legal advice or proposed medical treatments should be treated with due caution. will not be responsible for any consequences of actions that result from any form of use of any documents on this website.

Add your own feedback and questions here:
You are equally welcome to be positive or negative about any paper but please be polite. If you are being critical you must mention at least one specific error, otherwise your comment will be deleted as unhelpful.

comments powered by Disqus