Learning from the mistakes of others: matching errors in cross dataset learning

Sharmanska, Viktoriia and Quadrianto, Novi (2016) Learning from the mistakes of others: matching errors in cross dataset learning. Published in: Proceedings 29th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2016); Las Vegas, Nevada; 26 June - 1 July 2016. 3967-3975. Institute of Electrical and Electronics Engineers ISSN 1063-6919 ISBN 9781467388504

[img] PDF - Accepted Version
Download (2MB)

Abstract

Can we learn about object classes in images by looking at a collection of relevant 3D models? Or if we want to learn about human (inter-)actions in images, can we benefit from videos or abstract illustrations that show these actions? A common aspect of these settings is the availability of additional or privileged data that can be exploited at training time and that will not be available and not of interest at test time. We seek to generalize the learning with privileged information (LUPI) framework, which requires additional information to be defined per image, to the setting where additional information is a data collection about the task of interest. Our framework minimizes the distribution mismatch between errors made in images and in privileged data. The proposed method is tested on four publicly available datasets: Image+ClipArt, Image+3Dobject, and Image+Video. Experimental results reveal that our new LUPI paradigm naturally addresses the cross-dataset learning.

Item Type: Conference Proceedings
Schools and Departments: School of Engineering and Informatics > Informatics
Subjects: Q Science > QA Mathematics > QA0276 Mathematical statistics
Related URLs:
Depositing User: Novi Quadrianto
Date Deposited: 18 Apr 2016 09:26
Last Modified: 16 Jun 2017 10:21
URI: http://sro.sussex.ac.uk/id/eprint/60509

View download statistics for this item

📧 Request an update