Multi-feature combination generation and classification effectiveness evaluation using genetic algorithms
First Claim
1. A method for determining a preferred set of features for classifying content material, comprising:
- augmenting a pool of features with combination features that are based on two or more features in the pool of features, selecting a first plurality of feature sets from the pool of features, evaluating each feature set of the first plurality of feature sets to provide a measure of effectiveness of each feature set relative to each evaluated feature set'"'"'s ability to properly classify previously classified content material, and, selecting at least one subsequent plurality of feature sets from the pool of features based on the measure of effectiveness of each evaluated feature set, and evaluating each feature set of the at least one subsequent plurality of feature sets to provide the measure of effectiveness of each evaluated feature set relative to each feature set'"'"'s ability to properly classify previously classified content material, and, selecting the preferred set of features based on the measure of effectiveness of each evaluated feature set.
2 Assignments
0 Petitions
Accused Products
Abstract
The features that are presented to an evolutionary algorithm are preprocessed to generate combination features that may be more efficient in distinguishing among classifications than the individual features that comprise the combination feature. An initial set of features is defined that includes a large number of potential features, including the generated features that are combinations of other features. These features include, for example, all of the words used in a collection of content material that has been previously classified, as well as combination features based on these features, such as all the noun and verb phrases used. This pool of original features and combination features are provided to an evolutionary algorithm for a subsequent evaluation, generation, and determination of the best subset of features to use for classification. In this evaluation and generation process, each combination feature is processed as an independent feature, independent of the features that were used, or not used, to form the combination feature. In this manner, for example, a particular phrase that is generated as a combination of original feature words may be determined to be a better distinguishing feature than any of the original feature words and a more efficient distinguishing feature than an unrelated selection of the individual feature words, as might be provided by a conventional evolutionary algorithm. The resultant best performing subset is subsequently used to characterize new content material for automated classification. If the automated classification includes a learning system, the evolutionary algorithm and the generated combination features are also used to train the learning system.
44 Citations
12 Claims
-
1. A method for determining a preferred set of features for classifying content material, comprising:
-
augmenting a pool of features with combination features that are based on two or more features in the pool of features, selecting a first plurality of feature sets from the pool of features, evaluating each feature set of the first plurality of feature sets to provide a measure of effectiveness of each feature set relative to each evaluated feature set'"'"'s ability to properly classify previously classified content material, and, selecting at least one subsequent plurality of feature sets from the pool of features based on the measure of effectiveness of each evaluated feature set, and evaluating each feature set of the at least one subsequent plurality of feature sets to provide the measure of effectiveness of each evaluated feature set relative to each feature set'"'"'s ability to properly classify previously classified content material, and, selecting the preferred set of features based on the measure of effectiveness of each evaluated feature set. - View Dependent Claims (2, 3, 4, 5, 6)
-
-
7. A feature set selector that selects a preferred set of features that facilitates a classification of content material, comprising:
-
a feature combination generator that is configured to augment a pool of features by forming combination features that are based on two or more features of the pool of features, a subset selector that is configured to select a plurality of subsets of features from the pool of features, a classifier that is configured to classify prior classified content material using each subset of features of the plurality of subsets of features to provide an evaluation classification associated with each subset of features, an evaluator that is configured to evaluate each subset of features to provide a measure of effectiveness associated with each subset of features by comparing the evaluation classification associated with each subset of features and a proper classification associated with the prior classified content material, and an evolutionary algorithm that is configured to provide selection parameters based on the measure of effectiveness associated with each subset of features, and wherein the subset selector is configured to select subsequent pluralities of subsets of features based on the selection parameters, and the evolutionary algorithm provides the preferred set of features that facilitates the classification of content material in dependence upon the measure of effectiveness of each subset of features. - View Dependent Claims (8, 9)
-
-
10. A classification system comprising:
-
a feature combination generator that augments a pool of features with combination features that are a combination of other features in the pool of features, an evolutionary algorithm that is configured to generate sets of features based on measures of effectiveness associated with other sets of features, and to determine thereby a preferred set of features, and a classifier that is configured to classify content material based on the preferred set of features. - View Dependent Claims (11, 12)
-
Specification
-
- Resources
-
Current AssigneeKoninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
-
Original AssigneeKoninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
-
InventorsSchaffer, J. David
-
Primary Examiner(s)Knight, Anthony
-
Assistant Examiner(s)Holmes, Michael B.
-
Application NumberUS09/498,882Time in Patent Office1,919 DaysField of Search706 13- 15US Class Current706/14CPC Class CodesG06F 16/35 Clustering; ClassificationG06F 18/2111 by using evolutionary compu...G06N 3/126 Evolutionary algorithms, e....
