Unsupervised cross-domain distance metric adaptation with feature transfer network

1. A method for implementing an unsupervised cross-domain distance metric adaptation framework with a feature transfer network for enhancing facial recognition, comprising:

recursively training a feature transfer network based on a source domain associated with labeled source data and a target domain associated with unlabeled target data, and automatic labeling of target domain data using a clustering method, training the feature transfer network including:

training a first domain discriminator to discriminate between source-augmented features and target features and a second domain discriminator to discriminate between source features and a mixture of the source-augmented features and the target features, including training an objective function corresponding to the first domain discriminator and an objective function corresponding to the second domain discriminator; and

training a feature generator and a feature transformer based on the first and second domain discriminators, including training an objective function corresponding to the feature generator and an objective function corresponding to the feature transformer; and

implementing the feature transfer network and the automatic labeling to perform a facial recognition task.

2. The method of claim 1 , wherein:

training the feature generator includes verification loss from both the source features and the source-augmented features, domain separation loss via the second domain discriminator, and domain adversarial loss via the first domain discriminator; and

training the feature transformer includes verification loss and domain separation loss via the second domain discriminator.

3. The method of claim 1 , wherein training the feature generator and the feature transformer further includes training an objective function corresponding to verification loss associated with the feature generator, and training an objective function corresponding to verification loss associated with the feature transformer.

4. The method of claim 1 , wherein training the feature generator and the feature transformer further includes training an objective function corresponding to feature reconstruction loss between features extracted from the feature generator and a reference network pretrained using the labeled source data.

5. The method of claim 1 , wherein training the feature generator and the feature transformer further includes training an objective function corresponding to multi-class entropy minimization loss.

6. The method of claim 1 , wherein the automatic labeling includes implementing a hierarchical density-based spatial clustering of applications with noise method.

7. A computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method for implementing an unsupervised cross-domain distance metric adaptation framework with a feature transfer network for enhancing facial recognition, the method performed by the computer comprising:

recursively training a feature transfer network based on a source domain associated with labeled source data and a target domain associated with unlabeled target data, and automatic labeling of target domain data using a clustering method, training the feature transfer network including:

training a first domain discriminator to discriminate between source-augmented features and target features and a second domain discriminator to discriminate between source features and a mixture of the source-augmented features and the target features, including training an objective function corresponding to the first domain discriminator and an objective function corresponding to the second domain discriminator; and

training a feature generator and a feature transformer based on the first and second domain discriminators, including training an objective function corresponding to the feature generator and an objective function corresponding to the feature transformer; and

implementing the feature transfer network and the automatic labeling to perform a facial recognition task.

8. The computer program product of claim 7 , wherein:

training the feature generator includes verification loss from both the source features and the source-augmented features, domain separation loss via the second domain discriminator, and domain adversarial loss via the first domain discriminator; and

training the feature transformer includes verification loss and domain separation loss via the second domain discriminator.

9. The computer program product of claim 7 , wherein training the feature generator and the feature transformer further includes training an objective function corresponding to verification loss associated with the feature generator, and training an objective function corresponding to verification loss associated with the feature transformer.

10. The computer program product of claim 7 , wherein training the feature generator and the feature transformer further includes training an objective function corresponding to feature reconstruction loss between features extracted from the feature generator and a reference network pretrained using the labeled source data.

11. The computer program product of claim 7 , wherein training the feature generator and the feature transformer further includes training an objective function corresponding to multi-class entropy minimization loss.

12. The computer program product of claim 7 , wherein the automatic labeling includes implementing a hierarchical density-based spatial clustering of applications with noise method.

13. A system for implementing an unsupervised cross-domain distance metric adaptation framework with a feature transfer network for enhancing facial recognition, comprising:

a memory device for storing program code; and

at least one processor device operatively coupled to a memory device and configured to execute program code stored on the memory device to:

recursively train a feature transfer network based on a source domain associated with labeled source data and a target domain associated with unlabeled target data, and automatic labeling of target domain data using a clustering method, the at least one processor device being configured to train the feature transfer network by:

training a first domain discriminator to discriminate between source-augmented features and target features and a second domain discriminator to discriminate between source features and a mixture of the source-augmented features and the target features, including training an objective function corresponding to the first domain discriminator and an objective function corresponding to the second domain discriminator; and

training a feature generator and a feature transformer based on the first and second domain discriminators, including training an objective function corresponding to the feature generator and an objective function corresponding to the feature transformer; and

implement the feature transfer network and the automatic labeling to perform a facial recognition task.

14. The system of claim 13 , wherein:

the at least one processor device is further configured to train the feature generator by verification loss from both the source features and the source-augmented features, domain separation loss via the second domain discriminator, and domain adversarial loss via the first domain discriminator; and

the at least one processor device is further configured to train the feature transformer by verification loss and domain separation loss via the second domain discriminator.

15. The system of claim 13 , wherein the at least one processor device is further configured to train the feature generator and the feature transformer by training an objective function corresponding to verification loss associated with the feature generator, and training an objective function corresponding to verification loss associated with the feature transformer.

16. The system of claim 13 , wherein the at least one processor device is further configured to train the feature generator and the feature transformer by training an objective function corresponding to feature reconstruction loss between features extracted from the feature generator and a reference network pretrained using the labeled source data.

17. The system of claim 13 , wherein the at least one processor device is further configured to train the feature generator and the feature transformer by training an objective function corresponding to multi-class entropy minimization loss.

18. The system of claim 13 , wherein the automatic labeling includes implementing a hierarchical density-based spatial clustering of applications with noise method.