Dynamic model detecting apparatus
First Claim
1. A model detection apparatus for detecting a model from a series of input data, comprising:
- a plurality of estimate parameter memories for respectively storing mutually different distribution estimate parameters representing occurrences of said input data;
a plurality of distribution estimators respectively associated with said estimate parameter memories for producing a plurality of distribution estimate parameters from data stored in the associated estimate parameter memories and from said series of input data, and updating the associated estimate parameter memories with the produced distribution estimate parameters;
a model series memory for storing a plurality of candidate models corresponding in number to said estimate parameter memories;
a model series estimator for producing a plurality of candidate models from said series of input data, said distribution estimate parameters stored in said estimate parameter memories and said candidate models stored in said model series memory, and updating the model series memory with the produced candidate models; and
an optimal model series calculator for calculating an optimal series of models from said plurality of candidate models stored in said model series memory.
1 Assignment
0 Petitions
Accused Products
Abstract
A model detection apparatus comprises a number of estimate parameter memories for storing mutually different distribution estimate parameters representing occurrences of input data. A number of distribution estimators are respectively associated with the parameter memories for producing distribution estimate parameters from data stored in the associated parameter memories and from a series of input data, and updating the associated parameter memories with the produced parameters. A model series memory stores candidate models corresponding in number to the parameter memories. A model series estimator produces candidate models using the series of input data, the stored distribution estimate parameters and the stored candidate models, and updates the model series memory with the produced candidate models. An optimal model series calculator calculates an optimal series of models from the candidate models stored in the model series memory.
44 Citations
14 Claims
-
1. A model detection apparatus for detecting a model from a series of input data, comprising:
-
a plurality of estimate parameter memories for respectively storing mutually different distribution estimate parameters representing occurrences of said input data;
a plurality of distribution estimators respectively associated with said estimate parameter memories for producing a plurality of distribution estimate parameters from data stored in the associated estimate parameter memories and from said series of input data, and updating the associated estimate parameter memories with the produced distribution estimate parameters;
a model series memory for storing a plurality of candidate models corresponding in number to said estimate parameter memories;
a model series estimator for producing a plurality of candidate models from said series of input data, said distribution estimate parameters stored in said estimate parameter memories and said candidate models stored in said model series memory, and updating the model series memory with the produced candidate models; and
an optimal model series calculator for calculating an optimal series of models from said plurality of candidate models stored in said model series memory. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9)
-
-
10. A model detection apparatus for detecting a model from a series of input data, comprising:
-
a plurality of estimate parameter memories for respectively storing mutually different distribution estimate parameters representing occurrences of said input data;
a plurality of distribution estimators respectively associated with said estimate parameter memories for producing a plurality of distribution estimate parameters from data stored in the associated estimate parameter memories and from said series of input data, and updating the associated estimate parameter memories with the produced distribution estimate parameters;
a plurality of confidence calculators respectively associated with said estimate parameter memories for calculating confidence values of data stored in the associated estimate parameter memories;
a plurality of confidence memories for respectively storing the calculated confidence values;
a transition probability memory for storing a plurality of transition probability data;
a model transition probability estimator for estimating transition probability from data stored in said confidence memories and data stored in said transition probability memory and updating the transition probability memory with the estimated transition probability; and
an optimal model series calculator for calculating an optimal series of models from data stored in said confidence memories and data stored in said transition probability memory. - View Dependent Claims (11, 12, 13, 14)
-
Specification