To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
Stacked Support Vector Machine Ensembles for Cross-Culture Emotions Classification |
|
Author |
Khurshid Asghar, Mubbashar Sadddique, Inam ul Haq, M. Ahmad Nawaz-ul-Ghani, Ghulam Ali |
| Citation |
Vol. 19 No. 7 pp. 23-30 |
|
Abstract |
Facial expressions play a main role in representing the human’s internal emotional state in social communication, but the scope to which they are universal and culture reliant is a subject of discussion. In this paper, we introduced the Stacked Support Vector Machine Ensembles (SVMEs) for cross-culture emotions classification. A pool of SVM ensembles is stacked to learn the cross-culture emotions. The SVM ensemble is a collection of a set of support vector machines. The outcomes of support vector machines were tied to the probability distribution across the support vector machine ensembles. The final decision about the presence of an emotion is made by naive Bayes predictor. The cross-cultural facial images from JAFFE, TFEID, KDEF, CK+ and RadBoud databases are combined to develop the multi-culture dataset. The participants of multi-culture database originate from following geography and ethnicity: Japanese, Taiwanese, Caucasians, Moroccans, Swedish, Asians, Northern Europeans, Euro-American, and Afro-American. The experimental results and inter-expression resemblance analysis demonstrate that the proposed ensemble approach performs significantly better than the stat-of-the-art ensemble techniques. |
|
Keywords |
Support Vector Machine Ensemble, facial expression classification, boosted ensemble classifier, universal emotions classification |
|
URL |
