Detalle del congreso

Autores: P. Bulacio; E. Tapia; L. Angelone.

Resumen: The selection of informative and stable sets of genes with high recall values from real microarray datasets is a critical pre-processing task towards the correct classification of microarray data samples. The SVM-RFE algorithm is one of the best performing gene selection methods. However, the determination of the best filter-out policy for SVM-RFE is not trivial. Small filter-out factors tend to select small and unstable sets of genes with poor recall values at unaffordable computational costs. On the other hand, rough filter-out factors tend to select large and unstable sets of genes with acceptable recall values at affordable computational costs. The main goal of Sparse and Stable Consensus SVM-RFE gene selection (Sparse-S ) is the selection of small, informative and stable sets of genes with acceptable recall levels at affordable computational costs. Sparseness is achieved by the application of an AND constraint to gene sets attained with a number of diverse, but rough, SVM-RFE gene removal policies. After that, a stability constraint evaluates the gene selection frequency across multiple partitions of available data (e.g., 10-Fold CV). .

Tipo de reunión: Congreso.

Tipo de trabajo: Resumen.

Producción: Sparse and Stable Consensus SVM-RFE gene selection.

Reunión científica: II Congreso Argentino de Bioinformática y Biología Computacional.

Lugar: Córdoba.

Institución organizadora: Universidad Católica de Córdoba y Asociación Argentina de Bioinformática y Biología Computacional (A2B2C).

Publicado: Sí

Lugar publicación: Córdoba

Mes de reunión: 5

Año: 2011.

Página web: aquí