Method and apparatus for generating features through logical and functional operations

A system for generating features within a feature space, comprising:

• a feature generation and selection circuit within a hardware processor, the feature generation circuit configured to define an initial feature space for natural language processing of spoken input;
• split the initial feature space into a first plurality of feature sets using a dimension-based split strategy that splits the initial feature space into a plurality of disjoint feature sets, each feature set based on a feature dimension for the spoken input, wherein the feature dimensions comprise word, word position, speech tagger, and prosody characteristics, and further wherein the number of features in each feature set is defined in proportion to a relative importance of the respective dimension in a natural language processing application;
• execute a feature selection process on each feature set to select a first plurality of feature subsets from each of the first plurality of feature sets;
• merge the first plurality of feature subsets to produce a second feature space;
• split the second feature space into a second plurality of feature sets using the dimension-based split strategy, select a second plurality of feature subsets from each of the second plurality of feature subsets, and merge the second plurality of feature subsets to produce a third feature space;
• and
• perform at least one further feature selection process on the second feature space and any subsequent feature space using subsequent dimension-based split strategy and merging operations.

A non-transitory machine-readable medium including instructions which when executed in a processing system select an optimum feature subset from an initial feature set comprising:

• splitting the initial feature space into a first plurality of feature sets using a dimension-based split strategy that splits the initial feature space into a plurality of disjoint feature sets, each feature set based on a feature dimension for the spoken input, wherein the feature dimensions comprise word, word position, speech tagger, and prosody characteristics, and further wherein the number of features in each feature set is defined in proportion to a relative importance of the respective dimension in a natural language processing application;
• executing a feature selection process on each feature set to select a first plurality of feature subsets from each of the first plurality of feature sets;
• merging the first plurality of feature subsets to produce a second feature space;
• splitting the second feature space into a second plurality of feature sets using the dimension-based split strategy, selecting a second plurality of feature subsets from each of the second plurality of feature subsets, and merging the second plurality of feature subsets to produce a third feature space;
• and
• performing at least one further feature selection process on the second feature space if and any subsequent feature space using subsequent dimension-based split strategy and merging operations.