Device and method using machine learning model shared by plurality of applications
An electronic device may map a target application to a machine learning model matched to a request of the target application among a plurality of machine learning models, may generate an inference result for sensing data corresponding to the machine learning model based on the sensing data being sensed by the at least one sensor, and may transfer the generated inference result to at least one of the target application and another application.
1 . An electronic device comprising:
at least one processor comprising processing circuitry;
at least one sensor configured to sense sensing data,
memory comprising a storage device storing instructions, that when executed by the at least one processor individually or collectively, cause the electronic device to:
map a target application to a machine learning model matched to a request of the target application among a plurality of machine learning models;
based on sensing data corresponding to the machine learning model being sensed by the at least one sensor and the target application being deactivated, generate an inference result by applying the mapped machine learning model to the sensing data before the target application is activated;
activate the target application based on the inference result; and
transfer the generated inference result to the target application and another application mapped to the machine learning model.
2 . The electronic device of claim 1 , wherein the instructions, that when executed by the at least one processor individually or collectively, cause the electronic device to provide the generated inference result to the target application operating in one of a foreground process and a background process and the another application operating in one of the foreground process and the background process.
3 . The electronic device of claim 1 , the instructions, that when executed by the at least one processor individually or collectively, cause the electronic device to activate the deactivated target application based on at least one of an estimated label and an output value included in the inference result.
4 . The electronic device of claim 1 , the instructions, that when executed by the at least one processor individually or collectively, cause the electronic device to:
in response to an output value included in the inference result being out of a threshold range, activate the deactivated target application and transfer the inference result to the target application; and
in response to the output value included in the inference result being within the threshold range, maintain a state of the deactivated target application.
5 . The electronic device of claim 1 , the instructions, that when executed by the at least one processor individually or collectively, cause the electronic device to, in response to an estimated label included in the inference result being matched to a reference label registered in the deactivated target application, activate the deactivated target application.
6 . The electronic device of claim 1 , the instructions, that when executed by the at least one processor individually or collectively, cause the electronic device to, each time another inference result is generated based on the target application being activated, provide the another inference result to the target application.
7 . The electronic device of claim 1 , the instructions, that when executed by the at least one processor individually or collectively, cause the electronic device to determine the machine learning model matched to the target application, based on at least one or a combination of two or more of a type of input data required in an inference operation of the target application, a format of the input data, an input source, a type of output data, a label list of the output data, and a format of the output data, among the plurality of machine learning models.
8 . The electronic device of claim 1 , wherein: the instructions, that when executed by the at least one processor individually or collectively, cause the electronic device to generate the inference result by implementing the machine learning model on the sensing data, using an accelerator, and the accelerator comprises a at least one or a combination of two or more of a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), and a neural processing unit (NPU).
9 . The electronic device of claim 1 , wherein the plurality of machine learning models are stored in a memory space different from a memory space in which the target application is stored.
10 . A method implemented by at least one processor, the method comprising:
mapping a target application to a machine learning model matched to a request of the target application among a plurality of machine learning models;
based on sensing data corresponding to the machine learning model being sensed by at least one sensor and the target application being deactivated, generating an inference result by applying the mapped machine learning model to the sensing data before the target application is activated;
activating the target application based on the inference result; and
transferring the generated inference result to the target application and another application mapped to the machine learning model.
11 . The method of claim 10 , wherein the transferring of the generated inference result comprises providing the generated inference result to the target application operating in one of a foreground process and a background process and the another application operating in one of the foreground process and the background process.
12 . The method of claim 10 , wherein the activating of the target application comprises activating the deactivated target application based on at least one of an estimated label and an output value included in the inference result.
13 . The method of claim 10 , wherein the transferring of the generated inference result comprises:
in response to an output value included in the inference result being out of a threshold range, activating the deactivated target application and transferring the inference result to the target application; and
in response to the output value included in the inference result being within the threshold range, maintaining a state of the deactivated target application.
14 . The method of claim 10 , wherein the transferring of the generated inference result comprises, in response to an estimated label included in the inference result being matched to a reference label registered in the deactivated target application, activating the deactivated target application.
15 . The method of claim 10 , wherein the transferring of the generated inference result comprises, each time another inference result is generated based on the target application being activated, providing the another inference result to the target application.
16 . The method of claim 10 , wherein the mapping of the target application to the machine learning model comprises determining the machine learning model matched to the target application, based on at least one or a combination of two or more of a type of input data required in an inference operation of the target application, aformat of the input data, an input source, a type of output data, a label list of the output data, and a format of the output data, among the plurality of machine learning models.
17 . The method of claim 10 , wherein: the generating of the inference result comprises generating the inference result by implementing the machine learning model on the sensing data, using an accelerator, and the accelerator comprises a at least one or a combination of two or more of a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), and a neural processing unit (NPU).
18 . A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the operations of claim 10 .