ADAPTED DOMAIN SPECIFIC CLASS MEANS CLASSIFIER

• US 20170147944A1
• Filed: 11/24/2015
• Published: 05/25/2017
• Est. Priority Date: 11/24/2015
• Status: Active Grant

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Csurka, Gabriela, Chidlovskii, Boris, Clinchant, Stphane
• Granted Patent: US 10,296,846 B2
• Time in Patent Office: Days
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 3/02 Neural networks
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent

Abstract

A domain-adapted classification system and method are disclosed. The method includes mapping an input set of representations to generate an output set of representations, using a learned transformation. The input set of representations includes a set of target samples from a target domain. The input set also includes, for each of a plurality of source domains, a class representation for each of a plurality of classes. The class representations are representative of a respective set of source samples from the respective source domain labeled with a respective class. The output set of representations includes an adapted representation of each of the target samples and an adapted class representation for each of the classes for each of the source domains. A class label is predicted for at least one of the target samples based on the output set of representations and information based on the predicted class label is output.

Claims

1. A domain-adapted classification method comprising:

mapping an input set of representations to generate an output set of representations using a learned transformation, the input set of representations including a set of target samples from a target domain and, for each of a plurality of source domains, a class representation for each of a plurality of classes, the class representations each being representative of a set of source samples from the respective source domain labeled with a respective class, the output set of representations including an adapted representation of each of the target samples and an adapted class representation for each of the classes for each of the source domains;

predicting a class label for at least one of the target samples based on the output set of representations; and

outputting information based on the predicted class label,

wherein at least one of the mapping of the input set of representations and the predicting of the class label is performed with a processor.

2. The method of claim 1, wherein the class representations are class means.

3. The method of claim 1, wherein the mapping comprises, for at least one iteration:

learning a transformation that minimizes a reconstruction error when a corrupted set of representations generated from the input set is transformed, with the transformation, to generate a reconstructed set of representations from the input set of representations, and

outputting the reconstructed set of representations or adapted representations generated therefrom.

4. The method of claim 3, wherein the at least one iteration comprises a plurality of iterations and for a subsequent iteration, the input set is based on a reconstructed set of representations generated in a previous iteration.

5. The method of claim 4, wherein the input set for the subsequent iteration is generated by performing a non-linear transformation on the reconstructed set of representations generated in the previous iteration.

6. The method of claim 3, wherein the corrupted set of representations from the input set of representations corresponds to removing features from the input set with a predefined probability.

7. The method of claim 3, wherein for at least one iteration, the learning of the transformation is performed without generation of the corrupted set of representations, by using a closed form-solution to marginalize out noise.

8. The method of claim 7, wherein in the closed-form solution, the transformation

W=[P][Q]⁻

1, 

 

(4),

where the expectation of Q for a given entry in matrix [Q], denoted

9. The method of claim 1, wherein the predicting of the class label for the at least one of the target samples comprises computing one of:

a distance from the representation of the target sample in the output set to each of the adapted class representations in the output set; and

a distance from an augmented target representation to each of a set of augmented class representations, the augmented target representation being generated from the target representation in the output set, the augmented class representations being generated from the adapted class representations in the output set.

10. The method of claim 9, wherein the augmented representations are generated by at least one of:

concatenating the output sets of a plurality of the iterations; and

concatenating the input set and the output set.

11. The method of claim 1, wherein the predicting of the class label for the at least one of the target samples comprises, for each class, computing an optionally-weighted softmax distance from the adapted target representation to the adapted class representations for that class.

12. The method of claim 1, wherein the mapping and the prediction are performed without access to the source samples.

13. The method of claim 1, further comprising receiving the class representations for each of the plurality of source domains and the target samples and combining them to generate the input set of representations.

14. The method of claim 13, wherein the combining comprises concatenating the class representations and the target samples.

15. The method of claim 1, wherein the source samples and target samples are representations of objects selected from images and text documents.

16. The method of claim 1, wherein the outputting information comprises outputting a predicted class label for the at least one target sample.

17. The method of claim 1, wherein the class representations and target samples are each multidimensional representations comprising at least 10 dimensions.

18. A computer program product comprising a non-transitory recording medium storing instructions, which when executed on a computer, causes the computer to perform the method of claim 1.

19. A system comprising memory which stores instructions for performing the method of claim 1 and a processor in communication with the memory for executing the instructions.

20. A classification system comprising:

a mapping component which maps an input set of representations to generate an output set of representations, the input set of representations including a set of target samples from a target domain and, for each of a plurality of source domains, a class representation for each of a plurality of classes, the class representations each being representative of a set of source samples labeled with a respective class, the output set of representations including an adapted representation of each of the target samples and an adapted class representation for each of the classes for each of the source domains;

a classifier component which, for each of the classes, generates a classifier based on the adapted class representations for that class for each of the source domains and predicts a label for at least one of the target samples using the classifiers;

an output component which outputs information based on the predicted class label; and

a processor which implements the combining component, mapping component, classifier component, and output component.

21. The system of claim 20, wherein the output information comprises a predicted class label for the at least one target sample.

22. A classification method comprising:

mapping an input set of representations using a learned transformation to generate an output set of representations, the input set of representations including a set of target samples from a target domain and, for each of a plurality of source domains, a class representation for each of a plurality of classes, the class representations each being representative of a respective set of source samples from a respective one of the source domains labeled with a respective class, the output set of representations including an adapted representation of each of the target samples and an adapted class representation for each of the classes for each of the source domains;

for each class, generating a classifier with the adapted class representations for that class for each of the source domains;

predicting a class label for at least one of the target samples with the classifiers; and

outputting information based on the predicted class label,

wherein at least one of the mapping and the predicting is performed with a processor.