Method and apparatus for recognizing user based on on-device training

Patent metadata

Field	Value
Publication number	US-11900246-B2
Application number	US-202016914671-A
Country	US
Kind code	B2
Filing date	Jun 29, 2020
Priority date	Sep 2, 2019
Publication date	Feb 13, 2024
Grant date	Feb 13, 2024

Abstract

An on-device training-based user recognition method includes performing on-device training on a feature extractor based on reference data corresponding to generalized users and user data, determining a registration feature vector based on an output from the feature extractor in response to the input of the user data, determining a test feature vector based on an output from the feature extractor in response to an input of test data, and performing user recognition on a test user based on a result of comparing the registration feature vector to the test feature vector.

First claim

What is claimed is:

1. A recognition method comprising:
receiving user data input by a valid user for user registration;
performing on-device training on a feature extractor based on the user data and reference data corresponding to generalized users;
determining a registration feature vector based on an output from the feature extractor in response to input of the user data;
receiving test data input by a test user for user recognition;
determining a test feature vector based on an output from the feature extractor in response to input of the test data; and
performing the user recognition on the test user based on a result of comparing the registration feature vector to the test feature vector,
wherein the feature extractor includes a first neural network having a fixed parameter dependent on pretraining and a second neural network having an adjustable parameter dependent on the on-device training.

2. The recognition method of claim 1 , wherein the
wherein the adjustable parameter of the second neural network is adjusted by the on- device training.

3. The recognition method of claim 2 , wherein the first neural network is pretrained to extract a feature from input data based on a large user database.

4. The recognition method of claim 1 , wherein performing the on-device training comprises:
allocating labels of different values to the user data and the reference data, respectively; and
performing the on-device training based on a result of comparing the labels and outputs from the feature extractor in response to inputs of the user data and the reference data.

5. The method of claim 3 , further comprising performing the pretraining of the first neural network by a device different from a device that performs the user recognition.

6. The recognition method of claim 1 , wherein the reference data includes generalized feature vectors corresponding to the generalized users, and
wherein the generalized feature vectors are generated by grouping feature vectors corresponding to a plurality of generalized users into clusters.

7. The recognition method of claim 1 , wherein performing the user recognition comprises:
performing the user recognition based on a result of comparing a distance between the registration feature vector and the test feature vector to a threshold value.

8. The recognition method of claim 7 , wherein the distance between the registration feature vector and the test feature vector is determined based on one of a cosine distance between the registration feature vector and the test feature vector and an Euclidean distance between the registration feature vector and the test feature vector.

9. The recognition method of claim 1 , further comprising:
storing the determined registration feature vector in a registered user database.

10. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the recognition method of claim 1 .

11. A recognition apparatus comprising:
a processor; and
a memory including instructions executable in the processor, wherein, when the instructions are executed by the processor, the processor is configured to:
receive user data input by a valid user for user registration;
perform on-device training on a feature extractor based on the user data and reference data corresponding to generalized users;
determine a registration feature vector based on an output from the feature extractor in response to the input of the user data;
receive test data input by a test user for user recognition;
determine a test feature vector based on an output from the feature extractor in response to the input of the test data; and
perform the user recognition on the test user based on a result of comparing the registration feature vector to the test feature vector,
wherein the feature extractor includes a first neural network having a fixed parameter dependent on pretraining and a second neural network having an adjustable parameter dependent on an on-device training.

12. The recognition apparatus of claim 11 , wherein the adjustable parameter of the second neural network is adjusted by the on-device training.

13. The recognition apparatus of claim 12 , wherein the first neural network is pretrained to extract a feature from input data based on a large user database.

14. The recognition apparatus of claim 11 , wherein the processor is configured to:
allocate labels of different values to the user data and the reference data, respectively; and
perform the on-device training based on a result of comparing the labels and outputs from the feature extractor in response to inputs of the user data and the reference data.

15. The recognition apparatus of claim 11 , wherein the processor is configured to:
input the user data to the first neural network;
input, to the second neural network, the reference data and an output from the first neural network in response to the input of the user data; and
perform the on-device training based on an output from the second neural network.

16. The recognition apparatus of claim 11 , wherein the reference data includes generated feature vectors corresponding to the generalized users, and
wherein the generalized feature vectors are generated by grouping feature vectors corresponding to a plurality of generalized users into clusters.

17. The recognition apparatus of claim 11 , wherein the processor is configured to:
perform the user recognition based on a result of comparing a distance between the registration feature vector and the test feature vector to a threshold value.

18. The recognition apparatus of claim 17 , wherein the distance between the registration feature vector and the test feature vector is determined based on one of a cosine distance between the registration feature vector and the test feature vector and an Euclidean distance between the registration feature vector and the test feature vector.

19. The recognition apparatus of claim 11 , wherein the processor is configured to store the determined registration feature vector in a registered user database.

Assignees

• Samsung Electronics Co Ltd

Inventors

• Lee Dohwan
• Kim Kyuhong
• Han Jaejoon

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06V40/168

  Feature extraction; Face representation · CPC title

• G06V10/82

  using neural networks · CPC title

• G06N3/08 Primary

  Learning methods · CPC title