In this paper, we propose novel algorithmic models based on fusion of independent and correlated gene features for multiclass microarray gene expression classification. It is possible for genes to get co-expressed via different pathways. Moreover, a gene may or may not be co-active for all samples. In this paper, we approach this problem with a optimal feature selection technique using analysis based on statistical techniques to model the complex interactions between genes. The two different types of correlation modelling techniques based on the cross modal factor analysis (CFA) and canonical correlation analysis (CCA) were examined. The subsequent fusion of CCA/CFA features with principal component analysis (PCA) features at feature-level, and at score-level result in significant enhancement in classification accuracy for different data sets corresponding to multiclass microarray gene expression data.
|Title of host publication||Proceedings of the 13th International Conference on Information Fusion|
|Place of Publication||Piscataway, N.J., USA|
|Number of pages||6|
|Publication status||Published - 2010|
|Event||Fusion 2010: 13th International Conference on Information Fusion - Edinburgh, United Kingdom|
Duration: 26 Jul 2010 → 29 Jul 2010
|Conference||Fusion 2010: 13th International Conference on Information Fusion|
|Period||26/07/10 → 29/07/10|
Chetty, G., & Chetty, M. (2010). Multiclass microarray gene expression classification based on fusion of correlation features. In Proceedings of the 13th International Conference on Information Fusion (pp. 1-6). Piscataway, N.J., USA: IEEE.