Verimli‭ ‬yarı denetimli‭ ‬özellik seçimi‭ ‬için torbalama kısıtlamalı laplas skoru

In this study, we offer the efficiency and firmly approach for semi-supervised merit choices, depending on Constrained Laplacian Score (CLS). The main obstacle of this way is the options of the few choices in the supervision information, which is given by pairwise constraints. Actually, constraints are definite to calculate the noise which makes the learning performance may suffer from being the collapse. In this study we are trying to exceed any effects that effect on the performance of constraint set by the variation of their sources. This is done by technique staff using a resampling of data (bagging) and a random substance strategy. The high-dimensional datasets experiments are supported to confirm the proposed approximation, and comparing it with other representative advantage methods. Key words: Bagging, Constraint Score, Laplacian Score, Feature selection.