VideoSense at TRECVID 2011 : Semantic Indexing from Light Similarity Functions-based Domain Adaptation with Stacking

Emilie Morvant 1, 2, 3 Stéphane Ayache 1, 2 Amaury Habrard 3 Miriam Redi 4 Claudiu Tanase 4 Bernard Merialdo 4 Bahjat Safadi 5 Franck Thollard 6 Nadia Derbas 5 Georges Quénot 6, *
* Corresponding author
1 QARMA - éQuipe AppRentissage et MultimediA [Marseille]
LIF - Laboratoire d'informatique Fondamentale de Marseille
6 MRIM - Modélisation et Recherche d’Information Multimédia [Grenoble]
UPMF - Université Pierre Mendès France - Grenoble 2, UJF - Université Joseph Fourier - Grenoble 1, Inria - Institut National de Recherche en Informatique et en Automatique, INPG - Institut National Polytechnique de Grenoble , CNRS - Centre National de la Recherche Scientifique : UMR5217
Abstract : This paper describes our participation to the TRECVID 2011 challenge [1]. This year, we focused on a stacking fusion with Domain Adaptation algorithm. In machine learning, Domain Adaptation deals with learning tasks where the train and the test distributions are supposed related but different. We have implemented a classical approach for concept detection using individual features (low-level and intermediate features) and supervised classifiers. Then we combine the various classifiers with a second layer of classifier (stacking) which was specifically designed for Domain Adaptation. We show that, empirically, Domain Adaptation can improve concept detection by considering test information during the learning process.
Document type :
Conference papers
TRECVID 2011 - TREC Video Retrieval Evaluation workshop, Nov 2011, Gaithersburg, MD, United States. NIST, 6p., 2011


https://hal.archives-ouvertes.fr/hal-00685530
Contributor : Emilie Morvant <>
Submitted on : Thursday, April 5, 2012 - 3:13:23 PM
Last modification on : Tuesday, January 13, 2015 - 11:04:50 AM

File

videosense.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00685530, version 1

Citation

Emilie Morvant, Stéphane Ayache, Amaury Habrard, Miriam Redi, Claudiu Tanase, et al.. VideoSense at TRECVID 2011 : Semantic Indexing from Light Similarity Functions-based Domain Adaptation with Stacking. TRECVID 2011 - TREC Video Retrieval Evaluation workshop, Nov 2011, Gaithersburg, MD, United States. NIST, 6p., 2011. <hal-00685530>

Export

Share

Metrics

Record views

263

Document downloads

143