Fast microarray expression data analysis method for network exploration
AssigneeAgilent Technologies, Inc.
InventorsValery Kanevsky, Aditya Vailaya
Life of this patent
Estimating the term…
Abstract
A method for feature selection is provided. The method includes the steps of selecting a predictor set of features, adding at least one complementary feature to the predictor set based on a quality of prediction, checking to see if all of the features of the predictor set are repeated, and if not, removing at least one feature from the predictor set. The algorithm and method repeats the steps of adding complements, checking the predictor set and removing features until the features of the predictor set are repeated. Once the features of. the predictor set are repeated the proper number of times, the algorithm and method terminate.
Claims32
A method of feature selection comprising:
- (a) selecting a weighted subset of (k−x) features associated with a target, either individually or in combination, k being a number greater than 1 and x being a number greater than 0 and less than k;
- (b) adding x complementary features to said subset to form a predictor set of k features;
- (c) selecting a counting threshold, T, that is greater than or equal to k;
- (d) checking to see if all of said k features of said predictor set have been repeated T times in a row;
- (e) if determined in step (d) that all of said features of said predictor set have not been repeated T times in a row, removing at least x features from said predictor set and returning to step (b);
- and
- (f) as a result of determining in step (d) that all of said features of said predictor set have been repeated T times in a row, then selecting such predictor set as a best predictor set of k features for predicting said target.
A computer readable medium having computer executable instructions for feature selection that when executed cause a computer to perform the steps of:
- (a) selecting a weighted subset of (k−x) features associated with a target, either individually or in combination, k being a number greater than 1 and x being a number greater than 0 and less than k;
- (b) adding x complementary features to said subset to form a predictor set of k features;
- (c) means for selecting a counting threshold, T, that is greater than or equal to k;
- (d) checking to see if all of said k features of said predictor set have been repeated T times in a row;
- (e) if determined in step (d) that all of said features of said predictor set have not been repeated T times in a row, removing at least x features from said predictor set and returning to step (b);
- and
- (f) as a result of determining in step (d) that all of said features of said predictor set have been repeated T times in a row, then selecting such predictor set as a best predictor set of k features for predicting said target.
A method comprising the steps of:
- (a) selecting a subset of (k−1) features associated with a target;
- (b) adding at least one complement feature to said subset to form a predictor set of k features;
- (c) selecting a counting threshold, T, that is greater than or equal to k;
- (d) checking to see if all of said features of said predictor set have been repeated k times in a row;
- (e) if determined in step (d) that all of said features of said predictor set have not been repeated k times in a row, removing at least one feature from said predictor set and returning to step (b);
- and
- (f) as a result of determining if determined in step (d) that all of said features of said predictor set have been repeated k times in a row, then selecting such predictor set as a best predictor set of k features for predicting said target.
Dependencies are inferred from the claim text.