Method and apparatus for classification

1. A method for classification, comprising:

associating network data to be predicted with a predetermined number M of respective target classifiers, the predetermined number M being an integer that is greater than one;

obtaining a predicted result from each of the target classifiers;

obtaining a classification result of the network data based upon the predicted result and a prediction weight of each of the target classifiers;

determining a predetermined number N of candidate classifiers to be updated, the predetermined number N being an integer greater than or equal to the predetermined number M;

obtaining a third weight value for each of the candidate classifiers based upon a classification accuracy of each of the candidate classifiers;

obtaining a fourth weight value for each of the candidate classifiers based upon a second assigned time and construction time of each of the candidate classifiers; and

removing a predetermined number P of the candidate classifiers from the predetermined number N of the candidate classifiers to obtain the predetermined number M of the target classifiers, the predetermined number P being an integer that is greater than or equal to one and that is smaller than or equal to N−2,

wherein said determining, said obtaining the third weight value, said obtaining the fourth weight value and said removing each occur prior to said associating the network data.

2. The method of claim 1 , wherein each of the target classifiers is independent from others of the target classifiers.

3. The method of claim 1 , further comprising:

obtaining a first weight value of each of the target classifiers based upon a classification accuracy of each of the target classifiers;

obtaining a second weight value of each of the target classifiers based upon a first assigned time and construction time of each of the target classifiers; and

obtaining a prediction weight of each of the target classifiers based upon the first weight value and the second weight value,

wherein said obtaining the first weight value, said obtaining the second weight value and said obtaining the prediction weight each occur prior to said obtaining the classification result.

4. The method of claim 1 , further comprising using each training sample set of M training sample sets to construct one target classifier, wherein each of the training sample sets includes respective training samples that are not being identical among the training sample sets.

5. The method of claim 4 , wherein said using each training sample set occurs before said associating the network data.

6. The method of claim 1 , further comprising using one training sample set of M training sample sets to construct one target classifier using a respectively classification process selected from a group of M classification processes.

7. The method of claim 6 , wherein said using one training sample set occurs before said associating the network data.

8. The method of claim 1 , further comprising:

determining one constructed new candidate classifier;

obtaining a fifth weight value of the new candidate classifier based upon a classification accuracy of the new candidate classifier;

obtaining a sixth weight value for each of the candidate classifiers based upon a third assigned time and construction time of the new candidate classifier; and

identifying a predetermined number Q of the candidate classifiers to be updated and designating the new candidate classifier as the predetermined number M of the target classifiers based upon at least one of the fifth weight value and the sixth weight value,

wherein said determining, said obtaining the fifth weight value, said obtaining the sixth weight value, said identifying and said designating each occur prior to said associating the network data.

9. An apparatus for classification, comprising:

a classification unit configured to associate network data to be predicted with a predetermined number M of target classifiers respectively and for obtaining a predicted result from each of the target classifiers, each of the of target classifiers being independent from others of the target classifiers, the predetermined number M being an integer that is greater than one;

a processing unit configured to obtain a classification result of the network data based upon the predicted result and a prediction weight of each of the target classifiers; and

a first update unit configured to:

determine a predetermined number N of candidate classifiers to be updated, the predetermined number N being an integer greater than or equal to the predetermined number M;

obtain a third weight value for each of the candidate classifiers based upon a classification accuracy of each of the candidate classifiers;

obtain a fourth weight value for each of the candidate classifiers based upon a second assigned time and construction time of each of the candidate classifiers; and

remove a predetermined number P of the candidate classifiers from the predetermined number N of the candidate classifiers to obtain the predetermined number M of the target classifiers, the predetermined number P being an integer that is greater than or equal to one and that is smaller than or equal to N−2.

10. The apparatus of claim 9 , further comprising a construction unit configured to use each training sample set of M training sample sets to construct one target classifier, wherein each of the training sample sets includes respective training samples that are not being identical among the training sample sets.

11. The apparatus of claim 9 , wherein said processing unit is further configured to:

obtain a first weight value of each of the target classifiers based upon a classification accuracy of each of the target classifiers;

obtain a second weight value of each of the target classifiers based upon a first assigned time and construction time of each of the target classifiers; and

obtain a prediction weight of each of the target classifiers based upon the first weight value and the second weight value.

12. The apparatus of claim 9 , further comprises a second update unit configured to:

determine one constructed new candidate classifier;

obtain a fifth weight value of the new candidate classifier based upon a classification accuracy of the new candidate classifier;

obtain a sixth weight value for each of the candidate classifiers based upon a third assigned time and construction time of the new candidate classifier; and

identify a predetermined number Q of the candidate classifiers to be updated and instruction for designating the new candidate classifier as the predetermined number M of the target classifiers based upon at least one of the fifth weight value and the sixth weight value.

13. A non-transitory computer storage medium including at least one program for classification when implemented by a processor, comprising:

instruction for associating network data to be predicted with a predetermined number M of target classifiers respectively, each of the of target classifiers being independent from others of the target classifiers, the predetermined number M being an integer that is greater than one;

instruction for obtaining a predicted result from each of the target classifiers;

instruction for obtaining a classification result of the network data based upon the predicted result and a prediction weight of each of the target classifiers;

instruction for determining a predetermined number N of candidate classifiers to be updated, the predetermined number N being an integer greater than or equal to the predetermined number M;

instruction for obtaining a third weight value for each of the candidate classifiers based upon a classification accuracy of each of the candidate classifiers;

instruction for obtaining a fourth weight value for each of the candidate classifiers based upon a second assigned time and construction time of each of the candidate classifiers; and

instruction for removing a predetermined number P of the candidate classifiers from the predetermined number N of the candidate classifiers to obtain the predetermined number M of the target classifiers, the predetermined number P being an integer that is greater than or equal to one and that is smaller than or equal to N−2,

wherein said instruction for determining, said instruction for obtaining the third weight value, said instruction for obtaining the fourth weight value and said instruction for removing each occur prior to said instruction for associating the network data when implemented by the processor.

14. The computer storage medium of claim 13 , further comprising:

instruction for obtaining a first weight value of each of the target classifiers based upon a classification accuracy of each of the target classifiers;

instruction for obtaining a second weight value of each of the target classifiers based upon a first assigned time and construction time of each of the target classifiers; and

instruction for obtaining a prediction weight of each of the target classifiers based upon the first weight value and the second weight value,

wherein said instruction for obtaining the first weight value, said instruction for obtaining the second weight value and said instruction for obtaining the prediction weight each occur prior to said instruction for obtaining the classification result when implemented by the processor.

15. The computer storage medium of claim 13 , further comprising instruction for using each training sample set of M training sample sets to construct one target classifier,

wherein each of the training sample sets includes respective training samples that are not being identical among the training sample sets, and

wherein said instruction for using each training sample set occurs before said instruction for associating the network data when implemented by the processor.

16. The computer storage medium of claim 13 , further comprising:

instruction for determining one constructed new candidate classifier;

instruction for obtaining a fifth weight value of the new candidate classifier based upon a classification accuracy of the new candidate classifier;

instruction for obtaining a sixth weight value for each of the candidate classifiers based upon a third assigned time and construction time of the new candidate classifier; and

instruction for identifying a predetermined number Q of the candidate classifiers to be updated and instruction for designating the new candidate classifier as the predetermined number M of the target classifiers based upon at least one of the fifth weight value and the sixth weight value,

wherein said instruction for determining, said instruction for obtaining the fifth weight value, said instruction for obtaining the sixth weight value, said instruction for identifying and said instruction for designating each occur prior to said instruction for associating the network data when implemented by the processor.