PROJECT TITLE :

Learning Proximity Relations for Feature Selection

ABSTRACT:

This work presents a feature choice methodology based mostly on proximity relations learning. Each single feature is treated as a binary classifier that predicts for any three objects X, A, and B whether or not X is shut to A or B. The performance of the classifier may be a direct live of feature quality. Any linear combination of feature-based binary classifiers naturally corresponds to feature choice. So, the feature choice drawback is reworked into an ensemble learning problem of combining several weak classifiers into an optimized sturdy classifier. We provide a theoretical analysis of the generalization error of our proposed technique that validates the effectiveness of our proposed methodology. Varied experiments are conducted on synthetic data, four UCI knowledge sets and 12 microarray information sets, and demonstrate the success of our approach applying to feature selection. A weakness of our algorithm is high time complexity.