Mobile terminal and method of operating the same

Patent metadata

Field	Value
Publication number	US-10916240-B2
Application number	US-201916401726-A
Country	US
Kind code	B2
Filing date	May 2, 2019
Priority date	Dec 19, 2018
Publication date	Feb 9, 2021
Grant date	Feb 9, 2021

Abstract

A terminal includes a memory configured to store voice data and a processor configured to measure reliability of learnable data stored in the memory, to classify the learnable data into learning data or adaptive data according to the measured reliability, to generate a learning model by performing unsupervised learning with respect to the learning data, to generate an adaptive model using the adaptive data, and to evaluate recognition performance of each of the learning model and the adaptive model.

First claim

What is claimed is:

1. A terminal comprising:
a memory configured to store learnable data; and
a processor configured to:
measure reliability of the learnable data stored in the memory,
classify the learnable data as either learning data or adaptive data according to the measured reliability;
generate a learning model by performing unsupervised learning with respect to the learning data;
generate an adaptive model using the adaptive data; and
evaluate recognition performance of each of the learning model and the adaptive model.

2. The terminal of claim 1 , wherein the processor is further configured to:
classify the learnable data as the learning data if the measured reliability is equal to or greater than a reference reliability; and
classify the learnable data as the adaptive data if the measured reliability is less than the reference reliability.

3. The terminal of claim 2 , wherein the processor is further configured to:
select N data having a hidden Markov model-state entropy greater than a reference entropy from among the learning data having the measured reliability equal to or greater than the reference reliability;
perform unsupervised learning using the selected N data and prestored seed data; and
generate the learning model according to a result of performing the unsupervised learning.

4. The terminal of claim 2 , wherein the processor is further configured to:
generate the adaptive model using a generative adversarial network with respect to the adaptive data having the measured reliability less than the reference reliability.

5. The terminal of claim 1 , wherein the processor is further configured to:
measure a first performance evaluation value indicating a number of successes of voice recognition, if the learning model is applied to logging voice data;
measure a second performance evaluation value indicating a number of successes of voice recognition, if the adaptive model is applied to the logging voice data; and
select a model corresponding to a larger performance evaluation value of the first performance evaluation value and the second performance evaluation value of the learning model and the adaptive model.

6. The terminal of claim 5 , wherein the processor is further configured to:
compare a performance evaluation value of the selected model with that of a prestored acoustic model; and
update the acoustic model to the selected mode if the performance evaluation value of the selected model is greater than that of the acoustic model.

7. A method of operating a terminal having a memory, the method comprising:
measuring reliability of learnable data stored in the memory,
classifying the learnable data as either learning data or adaptive data according to the measured reliability;
generating a learning model by performing unsupervised learning with respect to the learning data;
generating an adaptive model using the adaptive data; and
evaluating recognition performance of each of the learning model and the adaptive model.

8. The method of claim 7 , wherein the classifying includes:
classifying the learnable data as the learning data if the measured reliability is equal to or greater than a reference reliability; and
classifying classify the learnable data as the adaptive data if the measured reliability is less than the reference reliability.

9. The method of claim 8 , wherein the generating of the learning model includes:
selecting N data having a hidden Markov model-state entropy greater than a reference entropy from among the learning data having the measured reliability equal to or greater than the reference reliability;
performing unsupervised learning using the selected N data and prestored seed data; and
generating the learning model according to a result of performing the unsupervised learning.

10. The method of claim 8 , wherein the generating of the adaptive model includes generating the adaptive model using a generative adversarial network with respect to the adaptive data having the measured reliability less than the reference reliability.

11. The method of claim 7 , wherein the evaluating of the recognition performance includes:
measuring a first performance evaluation value indicating a number of successes of voice recognition, if the learning model is applied to logging voice data;
measuring a second performance evaluation value indicating a number of successes of voice recognition, if the adaptive model is applied to the logging voice data; and
selecting a model corresponding to a larger performance evaluation value of the first performance evaluation value and the second performance evaluation value of the learning model and the adaptive model.

12. The method of claim 11 , further comprising:
comparing a performance evaluation value of the selected model with that of a prestored acoustic model; and
updating the acoustic model to the selected mode if the performance evaluation value of the selected model is greater than that of the acoustic model.

13. A non-transitory machine-readable medium comprising code, which, when executed by a machine, causes the machine to perform operations, the operations comprising:
measuring reliability of learnable data stored in memory,
classifying the learnable data as either learning data or adaptive data according to the measured reliability;
generating a learning model by performing unsupervised learning with respect to the learning data;
generating an adaptive model using the adaptive data; and
evaluating recognition performance of each of the learning model and the adaptive model.

Assignees

• Lg Electronics Inc

Inventors

• Lee Jeehye
• Yun Hwansik
• Kim Eugene

Classifications

• G06N3/047

  Probabilistic or stochastic networks · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N3/094

  Adversarial learning · CPC title

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title