Thinking of images as what they are: compound matrix regression for image classification

Ma, Zhigang, Yang, Yi, Nie, Feiping and Sebe, Nicu (2013). Thinking of images as what they are: compound matrix regression for image classification. In: IJCAI 2013 - Proceedings of the 23rd International Joint Conference on Artificial Intelligence. 23rd International Joint Conference on Artificial Intelligence, IJCAI 2013, Beijing, China, (1530-1536). 3-9 August 2013.

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
Author Ma, Zhigang
Yang, Yi
Nie, Feiping
Sebe, Nicu
Title of paper Thinking of images as what they are: compound matrix regression for image classification
Conference name 23rd International Joint Conference on Artificial Intelligence, IJCAI 2013
Conference location Beijing, China
Conference dates 3-9 August 2013
Proceedings title IJCAI 2013 - Proceedings of the 23rd International Joint Conference on Artificial Intelligence   Check publisher's open access policy
Place of Publication Menlo Park, CA, United States
Publisher AAAI Press / International Joint Conferences on Artificial Intelligence
Publication Year 2013
Sub-type Fully published paper
Open Access Status
ISBN 9781577356332
ISSN 1045-0823
Start page 1530
End page 1536
Total pages 7
Collection year 2014
Abstract/Summary In this paper, we propose a new classification framework for image matrices. The approach is realized by learning two groups of classification vectors for each dimension of the image matrices. One novelty is that we utilize compound regression models in the learning process, which endows the algorithm increased degree of freedom. On top of that, we extend the two-dimensional classification method to a semi-supervised classifier which leverages both labeled and unlabeled data. A fast iterative solution is then proposed to solve the objective function. The proposed method is evaluated by several different applications. The experimental results show that our method outperforms several classification approaches. In addition, we observe that our method attains respectable classification performance even when only few labeled training samples are provided. This advantage is especially desirable for real-world problems since precisely annotated images are scarce.
Q-Index Code E1
Q-Index Status Confirmed Code
Institutional Status UQ

 
Versions
Version Filter Type
Citation counts: Scopus Citation Count Cited 0 times in Scopus Article
Google Scholar Search Google Scholar
Created: Tue, 04 Mar 2014, 01:46:19 EST by System User on behalf of School of Information Technol and Elec Engineering