Class detection scheme and time mediated averaging of class dependent models
First Claim
Patent Images
1. A method for generating recognition models, the method comprising:
- receiving a first model based on a first set of training data, the first set of training data originating from a first set of common entities;
receiving a second model based on a second set of training data, the second set of training data originating from a second set of common entities;
determining a difference in model information between the first model and the second model; and
creating an independent model based on the first set of training data and the second set of training data if the difference in model information is insignificant.
3 Assignments
0 Petitions
Accused Products
Abstract
A method, system, and computer program for class detection and time mediated averaging of class dependent models. A technique is described to take advantage of gender information in training data and how obtain female, male, and gender independent models from this information. By using a probability value to average male and female Gaussian Mixture Models (GMMs), dramatic deterioration in cross gender decoding performance is avoided.
53 Citations
27 Claims
-
1. A method for generating recognition models, the method comprising:
-
receiving a first model based on a first set of training data, the first set of training data originating from a first set of common entities;
receiving a second model based on a second set of training data, the second set of training data originating from a second set of common entities;
determining a difference in model information between the first model and the second model; and
creating an independent model based on the first set of training data and the second set of training data if the difference in model information is insignificant. - View Dependent Claims (2, 3, 4, 5)
-
-
6. A system for generating recognition models, the method comprising:
-
a first model based on a first set of training data, the first set of training data originating from a first set of common entities;
a second model based on a second set of training data, the second set of training data originating from a second set of common entities; and
a processing module configured to create an independent model based on the first set of training data and the second set of training data if the difference in model information between first model and the second model is insignificant. - View Dependent Claims (7, 8, 9, 10)
-
-
11. A computer program product embodied in a tangible media comprising:
-
computer readable program codes coupled to the tangible media for generating recognition models, the computer readable program codes configured to cause the program to;
receive a first model based on a first set of training data, the first set of training data originating from a first set of common entities;
receive a second model based on a second set of training data, the second set of training data originating from a second set of common entities;
determine a difference in model information between the first model and the second model; and
create an independent model based on the first set of training data and the second set of training data if the difference in model information is insignificant. - View Dependent Claims (12, 13, 14, 15)
-
-
16. A system for generating recognition models, the method comprising:
-
a first model based on a first set of training data, the first set of training data originating from a first set of common entities;
a second model based on a second set of training data, the second set of training data originating from a second set of common entities; and
means for creating an independent model based on the first set of training data and the second set of training data if the difference in model information between first model and the second model is insignificant.
-
-
17. A method for recognizing data from a data stream originating from one of a plurality of data classes, the method comprising:
-
receiving a current feature vector;
computing a current vector probability that the current feature vector belongs to one of the plurality of data classes;
computing an accumulated confidence level that the data stream belongs to one of the plurality of data classes based on the current vector probability and on previous vector probabilities;
weighing class models based on the accumulated confidence; and
recognizing the current feature vector based on the weighted class models. - View Dependent Claims (18, 19, 20)
-
-
21. A system for recognizing data from a data stream originating from one of a plurality of data classes, the system comprising:
-
a receiving module configured to receive a current feature vector;
a first computing module configured to compute a current vector probability that the current feature vector belongs to one of the plurality of data classes;
a second computing module configured to compute an accumulated confidence level that the data stream belongs to one of the plurality of data classes based on the current vector probability and on previous vector probabilities;
a weighing module configured to weigh class models based on the accumulated confidence; and
a recognizing module configured to recognize the current feature vector based on the weighted class models. - View Dependent Claims (22, 23)
-
-
24. A computer program product embodied in a tangible media comprising:
-
computer readable program codes coupled to the tangible media for recognizing data from a data stream originating from one of a plurality of data classes, the computer readable program codes configured to cause the program to;
receive a current feature vector;
compute a current vector probability that the current feature vector belongs to one of the plurality of data classes;
compute an accumulated confidence level that the data stream belongs to one of the plurality of data classes based on the current vector probability and on previous vector probabilities;
weigh class models based on the accumulated confidence; and
recognize the current feature vector based on the weighted class models. - View Dependent Claims (25, 26, 27)
-
Specification