Conference paper Open Access

ONLINE MULTI-TASK LEARNING FOR SEMANTIC CONCEPT DETECTION IN VIDEO

Foteini Markatopoulou; Vasileios Mezaris; Ioannis Patras

In this paper we propose an online multi-task learning algorithm for video concept detection. In particular, we extend the Efficient Lifelong Learning Algorithm (ELLA) in the following ways: a) we solve the objective function of ELLA using quadratic programming instead of solving the Lasso problem, b) we add a new label-based constraint that considers concept correlations, c) we use linear SVMs as base learners instead of logistic regression. Experimental results show improvement over both the single-task learning methods typically used in this problem and the original ELLA algorithm.

Files (632.9 kB)
Name Size
icip16_1_preprint.pdf
md5:2ea63812f644dee948985e9a549b2177 		632.9 kB Download
55
37
views
downloads
Views 55
Downloads 37
Data volume 23.4 MB
Unique views 52
Unique downloads 37
More info on how stats are collected.
Indexed in
Publication date:
September 25, 2016
DOI:
10.1109/ICIP.2016.7532344

Zenodo DOI Badge

DOI 

10.1109/ICIP.2016.7532344

Markdown 

[![DOI](https://zenodo.org/badge/DOI/10.1109/ICIP.2016.7532344.svg)](https://doi.org/10.1109/ICIP.2016.7532344)

reStructedText 

.. image:: https://zenodo.org/badge/DOI/10.1109/ICIP.2016.7532344.svg
   :target: https://doi.org/10.1109/ICIP.2016.7532344

HTML 

<a href="https://doi.org/10.1109/ICIP.2016.7532344"><img src="https://zenodo.org/badge/DOI/10.1109/ICIP.2016.7532344.svg" alt="DOI"></a>

Image URL 

https://zenodo.org/badge/DOI/10.1109/ICIP.2016.7532344.svg

Target URL 

https://doi.org/10.1109/ICIP.2016.7532344

Keyword(s):
Concept detection Multi-task learning Video
Grants:
European Commission:
  • InVID - In Video Veritas – Verification of Social Media Video Content for the News Industry (687786)
Meeting:
IEEE International Conference on Image Processing (ICIP 2016), Phoenix Cinvention Center, Phoenix, Arizona 85004 USA, 25-28 September 2016
Communities:
License (for files):
Creative Commons Attribution 4.0 International

Share

Cite as

Export