Back to results
US 2002/0165839 A1Published application
Segmentation and construction of segmentation classifiers
OwnerKevin M. Taylor, Paul D. Whitneyinventors
Life of this patent
Estimating the term…
Abstract
Several approaches are provided for designing algorithms that allow for fast retrieval, classification, analysis or other processing of data, with minimal expert knowledge of the data being analyzed, and further, with minimal expert knowledge of the math and science involved in building classifications and performing other statistical data analysis. Further, methods of analyzing data are provided where the information being analyzed is not easily susceptible to quantitative description.
Claims48
A segmentation construction system to comprising:
- a segment select module arranged to interact with a plurality of data objects in a data set, and extract segments therefrom;
- a training module arranged to select and train at least one segment classifier based upon said segments extracted by said segment select module;
- an effectiveness module arranged to determine at least one performance measure for each segment classifier, wherein feedback is provided to direct refinement based upon said at least one performance measure from said effectiveness module to at least one of said segment select module to modify select ones of said segments, and said training module to modify select ones of said at least one segment classifier.
A computer based segmentation construction system comprising:
- a segment select module arranged to interact with a plurality of digitally stored data objects to extract segments therefrom;
- a classifier training module having:
- a classifier select module arranged to select a segment classifier set having at least one segment classifier;
- a training module arranged to train said segment classifier set based upon said segments;
- a classifier effectiveness module arranged to evaluate said segment classifier set and report classifier performance using at least one performance measure;
- a first feedback path from said classifier effectiveness module to said segment select module;
- and,
- a second feedback path from said classifier effectiveness module to said classifier training module, wherein said at least one performance measure directs whether feedback is required to said segment select module to modify said segments, to said classifier training module to modify said segment classifier set, or to both.
5 more elements
A segmentation construction system comprising:
- at least one processor;
- a storage device;
- an output device;
- and,
- software executable by said at least one processor for:
- accessing in said storage device digitally stored representations of data objects;
- extracting segments from said digitally stored representations of data objects;
- selecting at least one segment classifier defining a classifier set;
- training said at least one segment classifier using said segments;
- and,
- iteratively refining said at least one segment classifier until a predetermined stopping criterion is met, said at least one classifier refined by:
- providing a performance measure for at least one segment classifier;
- and,
- performing at least one of:
- extracting additional segments and training said at least one segment classifier thereon;
- modifying select ones of said segments and retraining said at least one segment classifier thereon;
- modifying said segment classifier set by either adding at least one new segment classifier or removing at least one segment classifier from said classifier set, wherein said classifier set is retrained on said segments;
- and,
- modifying at least one parameter of at least one segment classifier, wherein said at least one segment classifier is retrained;
- wherein said output device is adapted to output when said predetermined stopping criterion is met, at least one of a select one of said segment classifiers in said classifier set and a prepared data set that comprises segments extracted from said data objects and stored independently therefrom.
17 more elements
A segmentation construction system comprising:
- a storage device;
- an output device;
- and,
- a processor programmed to:
- access from said storage device, digitally stored representations of data objects;
- extract segments from said digitally stored representations of data objects;
- train a segment classifier set comprising at least one segment classifier using said segments;
- provide a performance measure for each of said at least one segment classifier;
- and,
- refine said segment classifier set based upon said performance measure by executing program code to perform at least one of a modification to said segments and modification to at least one of said classifiers, wherein said refinement continues iteratively until a stopping criterion is met, wherein said output device is adapted to output after said stopping criterion is met, at least one of a select one of said segment classifiers in said classifier set and a prepared data set that comprises segments extracted from said data objects and stored independently therefrom.
7 more elements
A computer readable carrier including a segmentation computer program that causes a computer to perform operations comprising:
- accessing from a storage device, digitally stored representations of data objects;
- identifying at least one measure of interest extracted from at least one of said data objects;
- extracting features from said at least on measure of interest;
- classifying said at least one measure of interest based upon said features using a segment classifier;
- determining at least one performance measure of the results of the classification;
- and,
- iteratively refining said segment classifier based upon said performance measure until a predetermined stopping criterion is met by performing for each iteration, at least one of:
- extracting additional segments from said data set, wherein said segment classifier is trained by said training process using said additional segments and a new performance measure of said classifier is recomputed by said effectiveness process;
- modifying at least one of said segments, wherein said segment classifier is retrained using the modified segments and a new performance measure of said segment classifier is recomputed;
- and,
- modifying said segment classifier, wherein the modified version of said classifier is retrained using said segments, and a new performance measure is recomputed.
8 more elements
A segmentation construction system comprising:
- a storage device;
- an output device;
- and,
- a processor programmed to:
- access from said storage device, digitally stored representations of data objects;
- identify at least one measure of interest extracted from at least one of said data objects;
- extract features from said at least on measure of interest;
- and,
- classify said at least one measure of interest based upon said features using a segment classifier;
- identify the results of the classification;
- and,
- iteratively refine said segmentation construction system based upon the identified results of the classification until a stopping criterion is met, the refinement arranged perform at least one operation to modify, add, and remove select ones of said at least one measure of interest.
9 more elements
A segmentation construction system comprising:
- a processor;
- a storage device;
- an output device;
- and,
- software executable by said processor for:
- accessing from said storage device, digitally stored representations of data objects;
- identifying at least one measure of interest extracted from at least one of said data objects;
- extracting features from said at least one measure of interest;
- classifying said at least one measure of interest based upon said features using a segment classifier;
- identifying the results of the classification;
- and,
- iteratively refining said segmentation construction system based upon the identified results of the classification until a stopping criterion is met, the refinement arranged perform at least one operation of modifying, adding, and removing select ones of said at least one measure of interest.
9 more elements
A method of performing segmentation comprising:
- integrating into a computer environment:
- a segment select module arranged to select segments from a data set;
- a segment extraction module arranged to extract segments from data objects;
- a classifier select module arranged to select at least one segment classifier;
- a classifier training module arranged to train said at least one segment classifier selected by said classifier select module based upon said segments;
- and,
- a classifier performance evaluation module arranged to report at least one performance measure for each segment classifier trained by said classifier training module;
- providing a data set comprising a plurality of data objects;
- using said segment select module to define segments of interest within said data set;
- using said segment extraction module to extract said segments;
- using said classifier select module to select at least one segment classifier;
- using said classifier training module to train said at least one candidate segment classifier using said segments;
- using said classifier performance evaluation module to report at least one performance measure for each segment classifier trained;
- and,
- using said report of said at least one performance measure to direct change to at least one of said segments and said at least one segment classifier.
12 more elements
Dependencies are inferred from the claim text.