Paper Info
Title
Semantic Concept Detection Using Weighted Discretization Multiple Correspondence Analysis for Disaster Information Management
Abstract
Multimedia semantic concept detection is an emerging research area in recent years. One of the prominent challenges in multimedia concept detection is data imbalance. In this study, a multimedia data mining framework for interesting concept detection in videos is presented. First, the Minimum Description Length (MDL) discretization algorithm is extended to handle the imbalanced data. Thereafter, a novel Weighted Discretization Multiple Correspondence Analysis (WD-MCA) algorithm based on the Multiple Correspondence Analysis (MCA) approach is proposed to maximize the correlation between the feature value pairs and concept classes by incorporating the discretization information captured from the MDL module. The proposed framework achieves promising performance to videos containing disaster events. The experimental results demonstrate the effectiveness of the WD-MCA algorithm, specifically for imbalanced datasets, compared to several existing methods.
Year
DOI
Venue
2016
10.1109/IRI.2016.82
2016 IEEE 17th International Conference on Information Reuse and Integration (IRI)
Keywords
Field
DocType
Weighted discretization,Multiple Correspondence Analysis (MCA),imbalanced data,video concept detection,disaster information management
Multiple correspondence analysis,Data mining,Discretization,Information management,Algorithm design,Visualization,Computer science,Minimum description length,Artificial intelligence,Machine learning,Semantics,Multimedia data mining
Conference
ISBN
Citations 
PageRank 
978-1-5090-3208-2
3
0.38
References 
Authors
29
2
Name
Order
Citations
PageRank
Samira Pouyanfar114113.06
Shu-Ching Chen21978182.74