SYSTEMS AND METHODS FOR DOMAIN ADAPTATION IN NEURAL NETWORKS USING DOMAIN CLASSIFIER

• US 20200134424A1
• Filed: 10/31/2018
• Published: 04/30/2020
• Est. Priority Date: 10/31/2018
• Status: Active Grant

• Current Assignee: Sony Interactive Entertainment Inc. (Sony Group Corp.)
• Original Assignee: Sony Interactive Entertainment Inc. (Sony Group Corp.)
• Inventors: Chen, Ruxin, Chen, Min-Hung, Yoo, Jaekwon, Liu, Xiaoyu
• Granted Patent: US 11,494,612 B2
• Time in Patent Office: Days
• CPC Class Codes: G06N 3/044 Recurrent networks, e.g. Hopfield networks
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent
• CPC Class Codes: G06V 10/82 using neural networks

Abstract

A domain adaptation module is used to optimize a first domain derived from a second domain using respective outputs from respective parallel hidden layers of the domains.

Claims

1. An apparatus, comprising:

at least one processor, and

at least one computer storage that is not a transitory signal and that comprises instructions executable by the at least one processor to;

access a first neural network, the first neural network being associated with a first data type;

access a second neural network, the second neural network being associated with a second data type different from the first data type;

provide, as input, first training data to the second neural network;

select a first layer, the first layer being a hidden layer of the second neural network;

identify an output from the first layer that was generated based on the first training data;

using a third neural network, determine whether the output from the first layer is from the first neural network, the third neural network being different from the first and second neural networks;

based on a determination that the output from the first layer is not from the first neural network, adjust one or more weights of the first layer.

2. The apparatus of claim 1, wherein the instructions are executable by the at least one processor to:

initially establish the second neural network by a copying of the first neural network.

3. The apparatus of claim 1, wherein the instructions are executable by the at least one processor to:

based on a determination that the output from the first layer is from the first neural network, decline to adjust one or more weights of the first layer.

4. The apparatus of claim 3, wherein the output is a first output, and wherein the instructions are executable by the at least one processor to:

based on a determination that the first output from the first layer is from the first neural network, select a second layer, the second layer also being a hidden layer of the second neural network;

identify a second output, the second output being from the second layer;

using the third neural network, determine whether the second output is from the first neural network; and

based on a determination that the second output is not from the first neural network, adjust one or more weights of the second layer.

5. The apparatus of claim 4, wherein the first and second layers of the second neural network are selected randomly.

6. The apparatus of claim 1, wherein the instructions are executable by the at least one processor to:

prior to using the third neural network to determine whether the output from the first layer is from the first neural network, adjust one or more weights of one or more layers of the third neural network so that the third neural network learns to correctly classify outputs from layers of either of the first neural network and the second neural network.

7. The apparatus of claim 6, wherein the third neural network operates in unsupervised mode to, using labeled data, learn to correctly classify outputs from layers of either of the first neural network and the second neural network.

8. A method, comprising:

accessing a first neural network, the first neural network being associated with a first data type;

accessing a second neural network, the second neural network being associated with a second data type different from the first data type;

providing, as input, first training data to the second neural network;

selecting a first layer, the first layer being a hidden layer of the second neural network;

identifying an output from the first layer that was generated based on the first training data;

using a third neural network, determining whether the output from the first layer is from the first neural network, the third neural network being different from the first and second neural networks;

based on determining that the output from the first layer is not from the first neural network, adjusting one or more weights of the first layer.

9. The method of claim 8, comprising:

determining, using the third neural network, whether the output from the first layer is from the first neural network at least in part by using the third neural network to identify the output from the first layer as pertaining to the first data type.

10. The method of claim 8, comprising:

based on determining that the output from the first layer is from the first neural network, declining to adjust one or more weights of the first layer.

11. The method of claim 10, wherein the output is a first output, and wherein the method comprises:

based on determining that the first output from the first layer is from the first neural network, selecting a second layer, the second layer also being a hidden layer of the second neural network;

identifying a second output from the second layer;

using the third neural network, determining whether the second output is from the first neural network; and

based on determining that the second output is not from the first neural network, adjusting one or more weights of the second layer.

12. The method of claim 8, wherein the first layer is selected based on a command from a human supervisor.

13. The method of claim 8, wherein the instructions are executable by the at least one processor to:

prior to using the third neural network to determine whether the output from the first layer is from the first neural network, adjusting one or more weights of one or more layers of the third neural network so that the third neural network learns to correctly classify outputs from layers of either of the first neural network and the second neural network.

14. The method of claim 8, wherein the third neural network operates in unsupervised mode to, using labeled data, learn to correctly classify outputs from layers of either of the first neural network and the second neural network.

15. The method of claim 8, comprising:

initially establishing the second neural network by a copying of the first neural network.

16. An apparatus, comprising:

at least one computer storage that is not a transitory signal and that comprises instructions executable by at least one processor to;

access a first domain, the first domain being associated with a first domain genre;

access a second domain, the second domain being associated with a second domain genre different from the first domain genre;

using training data provided to the first and second domains, classify a target data set; and

output a classification of the target data set, wherein the target data set is classified by a domain adaptation module comprising a domain classifier to inverse a gradient and back-propagate the gradient to a main model.

17. The apparatus of claim 16, wherein the first domain pertains to real world video and the second domain pertains to computer game video.

18. The apparatus of claim 16, wherein the first domain pertains to information derived from a first voice and the second domain pertains to information derived from a second voice.

19. The apparatus of claim 16, wherein the first domain pertains to standard font text and the second domain pertains to cursive script.

20. The apparatus of claim 16, wherein the domain classifier inverses the gradient using a gradient reversal layer (GRL) receiving data from a spatial model and a temporal model.