Artificial intelligence apparatus and method for updating artificial intelligence model

[0000]
Disclosed herein an artificial intelligence apparatus for updating an artificial intelligence model including a memory configured to store an artificial intelligence model and training data for the artificial intelligence model and a processor configured to receive sensor data, calculate a confidence level of the received sensor data for the stored artificial intelligence model, label the received sensor data if the calculated confidence level is less than a reference value, store the labeled received sensor data in the memory as the training data or test data, and update the stored artificial intelligence model using the stored training data.

[00000]

1. An artificial intelligence apparatus for updating an artificial intelligence model, the artificial intelligence apparatus comprising:

a memory configured to store an artificial intelligence model and training data for the artificial intelligence model; and

a processor configured to:

receive sensor data,

calculate a confidence level of the received sensor data for the stored artificial intelligence model,

label first sensor data when the calculated confidence level is less than a reference value, and exclude from labeling, second sensor data when the calculated confidence level is equal to or greater than the reference value,

store the labeled first sensor data in the memory as the training data or test data, and

update the stored artificial intelligence model using only the stored training data.

2. The artificial intelligence apparatus of
claim 1
, wherein the processor is configured to label the first sensor data by extracting a pseudo label from the received sensor data using a pseudo labeler.

3. The artificial intelligence apparatus of
claim 2
, wherein the pseudo labeler has the same model parameter as the stored artificial intelligence model.

4. The artificial intelligence apparatus of
claim 1
, wherein the processor is configured to update model parameters of the artificial intelligence model stored in the memory using the training data stored in the memory.

5. The artificial intelligence apparatus of
claim 4
, further comprising a learning processor,

wherein the processor is configured to update the artificial intelligence model stored in the memory using the learning processor.

6. The artificial intelligence apparatus of
claim 4
, wherein the processor is configured to evaluate performance of the updated artificial intelligence model using the test data stored in the memory.

7. The artificial intelligence apparatus of
claim 1
, wherein the processor is configured to:

generate a result corresponding to the received sensor data using the stored artificial intelligence model, and

calculate the confidence level of the received sensor data based on the generated result.

8. The artificial intelligence apparatus of
claim 7
, wherein the confidence level of the received sensor data is determined based on an entropy corresponding to the generated result, a first-rank confidence level included in the generated result or a difference between the first-rank confidence level and a second-rank confidence level included in the generated result.

9. The artificial intelligence apparatus of
claim 8
, wherein the processor is configured to:

calculate the confidence level to be higher as the entropy decreases,

calculate the confidence level to be higher as the first-rank confidence level increases, and

calculate the confidence level to be higher as the difference between the first-rank confidence level and the second-rank confidence level increases.

10. The artificial intelligence apparatus of
claim 1
, wherein the sensor data includes at least one of sound data, image data, text data or measured data.

11. The artificial intelligence apparatus of
claim 10
, further comprising a sensor device including at least one sensor,

wherein the processor is configured to receive the sensor data from the sensor device.

12. The artificial intelligence apparatus of
claim 11
, further comprising a communicator configured to communicate with an external device, wherein the processor is configured to receive the sensor data from the external device via the communicator.

13. A method of updating an artificial intelligence model, the method performed by an artificial intelligence apparatus, the method comprising:

receiving sensor data,

calculating a confidence level of the received sensor data for an artificial intelligence model stored in a memory,

labeling first sensor data when the calculated confidence level is less than a reference value, and excluding from labeling, second sensor data when the calculated confidence level is equal to or greater than the reference value,

storing the labeled first sensor data in the memory as training data or test data, and

updating the stored artificial intelligence model using only the stored training data.

14. A non-transitory computer readable recording medium containing a program for causing an artificial intelligence apparatus, the method comprising:

receiving sensor data,

calculating a confidence level of the received sensor data for an artificial intelligence model stored in a memory,

labeling first sensor data when the calculated confidence level is less than a reference value, and excluding from labeling, second sensor data where the calculated confidence level is equal to or greater than the reference value,

storing the labeled first sensor data in the memory as training data or test data, and updating the stored artificial intelligence model using only the stored training data.

CROSS-REFERENCE TO RELATED APPLICATIONS
[0001]
This application claims priority to Korean Patent Application No. 10-2019-0111340 filed on Sep. 9, 2019, the entire contents of which is hereby incorporated by reference in its entirety.

BACKGROUND
[0002]
The present disclosure relates to an artificial intelligence (AI) apparatus and method for updating an artificial intelligence model, and more particularly, to an artificial intelligence apparatus and method for directly selecting training data to be used to train an artificial intelligence model and updating the artificial intelligence model.

[0003]
Recently, apparatuses for providing various functions using an artificial intelligence model generated using a machine learning algorithm or a deep learning algorithm are increasing. For example, apparatuses for interacting with users by speech using a speech recognition model or recognizing an object or a user using an image recognition model are increasing.

[0004]
Currently, due to limitations in storage space or computing power, most edge devices do not have artificial intelligence models installed therein. Similarly, most edge devices cannot directly train the artificial intelligence models due to limitations in computing power of the edge devices. Such edge devices operate depending on an artificial intelligence (AI) server and communication with the AI server is essential.

[0005]
In addition, a lot of human resources are required to generate a label in training data used to train the artificial intelligence model.

[0006]
As the storage space and computing power of the edge devices increase with technical advancement, it is expected that the edge devices will directly train, store and use the artificial intelligence models. However, since the performance of the edge devices is still inferior to the AI server, there is a need for technique for efficiently training the artificial intelligence model without human intervention.

SUMMARY
[0007]
An object of the present disclosure is to provide an artificial intelligence apparatus and method for storing an artificial intelligence model, selecting sensor data to be used to train the artificial intelligence model from acquired sensor data to generate training data, and updating the artificial intelligence model using the generated training data.

[0008]
According to an embodiment, provided are an artificial intelligence apparatus for receiving sensor data, calculating a confidence level of the received sensor data for a stored artificial intelligence model, labeling sensor data, the calculated confidence level of which is less than a reference value, storing the labeled sensor data as the training data or test data, and updating the stored artificial intelligence model using the stored training data, and a method thereof.

[0009]
According to an embodiment, provided are an artificial intelligence apparatus for labeling sensor data by extracting a label from sensor data, the calculated confidence level of which is less than a reference value, using a pseudo labeler and a method thereof.

[0010]
According to an embodiment, provided are an artificial intelligence apparatus for generating a result from received sensor data using a stored artificial intelligence model and calculating a confidence level of the received sensor data based on an entropy corresponding to the generated result, a first-rank confidence level included in the generated result or a difference between the first-rank confidence level and a second-rank confidence level included in the generated result, and a method thereof.

BRIEF DESCRIPTION OF THE DRAWINGS
[0011]
The present disclosure will become more fully understood from the detailed description given herein below and the accompanying drawings, which are given by illustration only, and thus are not limitative of the present disclosure, and wherein:

[0012]
FIG. 1 is a block diagram illustrating an AI apparatus according to an embodiment of the present disclosure;

[0013]
FIG. 2 is a block diagram illustrating an AI server according to an embodiment of the present disclosure;

[0014]
FIG. 3 is a view illustrating an AI system according to an embodiment of the present disclosure;

[0015]
FIG. 4 is a block diagram illustrating an AI apparatus according to an embodiment of the present disclosure;

[0016]
FIG. 5 is a flowchart illustrating a method of updating an artificial intelligence model according to an embodiment of the present disclosure;

[0017]
FIG. 6 is a view illustrating a method of updating an artificial intelligence model according to an embodiment of the present disclosure;

[0018]
FIG. 7 is a view illustrating an example of updating an artificial intelligence model according to an embodiment of the present disclosure; and

[0019]
FIG. 8 is a view illustrating an example of updating an artificial intelligence model according to an embodiment of the present disclosure.

DETAILED DESCRIPTION OF THE EMBODIMENTS
[0020]
Hereinafter, embodiments of the present disclosure are described in more detail with reference to accompanying drawings and regardless of the drawings symbols, same or similar components are assigned with the same reference numerals and thus overlapping descriptions for those are omitted. The suffixes “module” and “unit” for components used in the description below are assigned or mixed in consideration of easiness in writing the specification and do not have distinctive meanings or roles by themselves. In the following description, detailed descriptions of well-known functions or constructions will be omitted since they would obscure the disclosure in unnecessary detail. Additionally, the accompanying drawings are used to help easily understanding embodiments disclosed herein but the technical idea of the present disclosure is not limited thereto. It should be understood that all of variations, equivalents or substitutes contained in the concept and technical scope of the present disclosure are also included.

[0021]
It will be understood that the terms “first” and “second” are used herein to describe various components but these components should not be limited by these terms. These terms are used only to distinguish one component from other components.

[0022]
In this disclosure below, when one part (or element, device, etc.) is referred to as being ‘connected’ to another part (or element, device, etc.), it should be understood that the former can be ‘directly connected’ to the latter, or ‘electrically connected’ to the latter via an intervening part (or element, device, etc.). It will be further understood that when one component is referred to as being ‘directly connected’ or ‘directly linked’ to another component, it means that no intervening component is present.

[0023]

[0024]
Artificial intelligence refers to the field of studying artificial intelligence or methodology for making artificial intelligence, and machine learning refers to the field of defining various issues …