Selection of high risk patients with ranked models based on the CPL criterion functions
ICDM'10 Proceedings of the 10th industrial conference on Advances in data mining: applications and theoretical aspects
Prognostic models based on linear separability
ICDM'11 Proceedings of the 11th international conference on Advances in data mining: applications and theoretical aspects
Prognostic modeling with high dimensional and censored data
ICDM'12 Proceedings of the 12th Industrial conference on Advances in Data Mining: applications and theoretical aspects
| Hi-index | 0.00 |
Ranked models in the form of linear transformations of multivariate feature vectors on a line can be found on the basis of a priori given order within particular pairs of objects or events. Such ranked transformations are designed to preserve given sequential order. In this way, the sequential patterns inside sets of the feature vectors can be discovered and modelled. Attention is paid here to combining problems of sequential patterns modelling and recognition with feature selection. The feature selection problem is aimed at the best representation of the sequential patterns. The convex and piecewise linear (CPL) criterion functions are used here both for designing ranked linear models and for feature selection.