Solving imbalanced dataset problems for high-dimensional image processing by swarm optimization
Li J.; Fong S.
AbstractIn this chapter, techniques used to optimize the imbalanced class and high-dimensional image datasets using swarm intelligence (SI) algorithms are proposed. Datasets converted from images or multimedia usually have problems of imbalanced class distribution and high-dimensional features. These problems seriously affect the accuracy and efficiency of image processing, especially in machine learning. Compared with other methods, SI optimization algorithms can simultaneously and stochastically solve these two problems in a search space. The SI optimization algorithm is a relatively new approach in the field of artificial intelligence. Specifically. the classical particle swarm optimization and contemporary bat-inspired algorithm are adopted in our experiment. Our proposed method achieved high reliability and high accuracy in classification performance from a computer simulation experiment. Moreover, it can synthesize more reasonable minority class samples as well as select the appropriate features when compared to the existing methods.
KeywordFeature selection Imbalanced dataset Multimedia and image dataset Optimization Swarm intelligence algorithm
URLView the original
Fulltext Access
Citation statistics
Document TypeBook
CollectionUniversity of Macau
AffiliationUniversidade de Macau
Recommended Citation
GB/T 7714
Li J.,Fong S.. Solving imbalanced dataset problems for high-dimensional image processing by swarm optimization[M],2016.
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Li J.]'s Articles
[Fong S.]'s Articles
Baidu academic
Similar articles in Baidu academic
[Li J.]'s Articles
[Fong S.]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Li J.]'s Articles
[Fong S.]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.