Automatic feature selection system for data containing missing values
AssigneeThe United States of America as represented by the Secretary of the Navy
InventorsPeter K. Willett, Robert S. Lynch, Jr.
Life of this patent
Estimating the term…
Abstract
An automatic feature selection system for test data with data (including the test data and/or the training data containing missing values in order to improve classifier performance. The missing features for such data are selected in one of two ways: first approach assumes each missing feature is uniformly distributed over its range of values whereas in the second approach, the number of discrete levels for each feature is increased by one for the missing features. These two choices modify the Bayesian Data Reduction Algorithm accordingly used for the automatic feature selection.
Claims8
A data reduction method for a classification system implemented as software within a computer comprising the steps of:
- receiving training data in the classification system;
- processing said received training data to account for missing values in said received training data;
- forming quantized feature vectors for each class of the classification system from said processed training data, each quantized feature vector having a plurality of features, and each feature having data mapped into levels;
- applying a Bayesian data reduction algorithm to said classification system having quantized feature vectors to develop reduced feature vectors for said classification system;
- receiving test data;
- processing said received test data to account for missing values in said received test data;
- quantizing said processed test data into said reduced feature vectors;
- and
- testing said classification system having said reduced feature vectors using said quantized test data in order to provide tested quantized test data.
A data reduction method for a classification system implemented as software within a computer comprising the steps of:
- receiving training data in the classification system;
- forming quantized feature vectors for each class of the classification system from said received training data, each quantized feature vector having a plurality of features, and each feature having data mapped into levels;
- applying a Bayesian data reduction algorithm to said classification system having quantized feature vectors to develop reduced feature vectors for said classification system;
- receiving test data;
- processing said received test data to account for missing values in said received test data;
- quantizing said processed test data into said reduced feature vectors;
- and
- testing said classification system having said reduced feature vectors using said quantized test data in order to provide tested quantized test data.
Dependencies are inferred from the claim text.