Classifier training using noisy samples
An example system includes a processor to receive input data comprising noisy positive data and clean negative data. The processor is to cluster the input data. The processor is to compute a potential score for each cluster of the clustered input data. The processor is to iteratively refine cluster quality of the clusters using the potential scores of the clusters as weights. The processor is to train a classifier by sampling the negative dataset uniformly and the positive set in a non-uniform manner based on the potential score.
1. A system, comprising a processor to:
receive input data comprising noisy positive data and clean negative data;
cluster the input data;
compute a potential score for each cluster of the clustered input data;
iteratively refine cluster quality of the clusters using the potential scores of the clusters as weights; and
train a classifier by sampling the negative dataset uniformly and the positive set in a non-uniform manner based on the potential score.
2. The system of claim 1 , wherein the potential score is based on a positive ratio and a cluster distance variance.
3. The system of claim 1 , wherein the noisy positive data comprises a noise level of more than 50%.
4. The system of claim 1 , wherein the processor is to iteratively refine the clusters using a weighted k-means.
5. The system of claim 1 , wherein the processor is to train the classifier using the potential score as a distribution.
6. A computer-implemented method, comprising:
receiving, via a processor, input data comprising noisy positive data and clean negative data;
clustering, via the processor, the input data;
computing, via the processor, a potential score for each cluster of the clustered input data;
iteratively refining, via the processor, a cluster quality of the clusters using the potential scores of the clusters as weights; and
training, via the processor, a classifier by sampling the negative dataset uniformly and the positive set in a non-uniform manner based on the potential score.
7. The computer-implemented method of claim 6 , comprising computing the potential score based on a positive ratio and a cluster distance variance.
8. The computer-implemented method of claim 6 , wherein clustering the input data comprises performing a weighted deep embedded clustering.
9. The computer-implemented method of claim 6 , wherein iteratively refining the clusters comprises refining the clusters using a weighted k-means.
10. The computer-implemented method of claim 6 , wherein training the classifier comprises using the potential score as a distribution.
11. A computer-implemented method, comprising:
receiving, via a processor, data to be classified;
classifying, via a classifier trained via high noise training, using noisy positive data and clean negative data the data.
12. The computer-implemented method of claim 11 , wherein classifying the data comprises classifying the data with a positive label or a negative label.
13. The computer-implemented method of claim 11 , comprising training, via the processor, the classifier using iteratively refined clusters of noisy positive samples.
14. The computer-implemented method of claim 11 , comprising displaying the classified data.
15. A computer program product for training classifiers, the computer program product comprising a computer-readable storage medium having program code embodied therewith, wherein the computer readable storage medium is not a transitory signal per se, the program code executable by a processor to cause the processor to:
receive input data comprising noisy positive data and clean negative data;
cluster the input data;
compute a potential score for each cluster of the clustered input data;
iteratively refine cluster quality of the clusters using the potential scores of the clusters as weights; and
train a classifier by sampling the negative dataset uniformly and the positive set in a non-uniform manner based on the potential score.
16. The computer program product of claim 15 , further comprising program code executable by the processor to compute the potential score based on a positive ratio.
17. The computer program product of claim 15 , further comprising program code executable by the processor to compute the potential score based on a cluster distance variance.
18. The computer program product of claim 15 , further comprising program code executable by the processor to compute the potential score based on a positive ratio and a cluster distance variance.
19. The computer program product of claim 15 , further comprising program code executable by the processor to iteratively refine the clusters using a weighted k-means.
20. A computer program product for classifying data, the computer program product comprising a computer-readable storage medium having program code embodied therewith, wherein the computer readable storage medium is not a transitory signal per se, the program code executable by a processor to cause the processor to:
receive data to be classified; and
classify the data via a classifier trained via high noise training using noisy positive data and clean negative data.
21. The computer program product of claim 20 , further comprising program code executable by the processor to classify the data with a positive label or a negative label.
22. The computer program product of claim 20 , further comprising program code executable by the processor to train the classifier using iteratively refined clusters of noisy positive samples.
23. The computer program product of claim 20 , further comprising program code executable by the processor to display the classified data.