Model management device and model managing method
The model management device includes a communication unit capable of communicating with a plurality of AI devices, each of which continuously performs training of a machine learning model, and a processor configured to manage training of a machine learning model in each of the plurality of AI devices, and acquire accuracy of the machine learning model trained in each of the plurality of AI devices. The processor is configured to change an execution frequency of a process relating to training of the machine learning model having an accuracy within a predetermined range.
1 . A model management device for managing a plurality of machine learning models, comprising:
a communication unit capable of communicating with a plurality of AI devices, each of which continuously performs training of a machine learning model; and
a processor configured to manage training of the machine learning model in each of the plurality of AI devices, acquire an accuracy of the machine learning model trained in each of the plurality of AI devices, and acquire an amount of power that can be supplied to the plurality of AI devices, wherein
the processor is configured to change an execution frequency of acquiring data relating to training of the machine learning model having the accuracy within a predetermined range, based on the amount of power.
2 . The model management device as claimed in claim 1 , wherein the processor is configured to increase the execution frequency of acquiring the data related to training of the machine learning model having the accuracy equal to or less than a predetermined lower threshold.
3 . The model management device as claimed in claim 2 , wherein the processor is configured to restore the execution frequency to an original value when the accuracy of the machine learning model rises to a predetermined first determination value higher than the lower threshold due to the increase in the execution frequency.
4 . The model management device as claimed in claim 1 , wherein the processor is configured to reduce the execution frequency of acquiring the data related to training of the machine learning model having the accuracy equal to or higher than a predetermined upper threshold.
5 . The model management device as claimed in claim 4 , wherein the processor is configured to restore the execution frequency to an original value when the accuracy of the machine learning model drops to a predetermined second determination value less than the upper threshold due to the reduction of the execution frequency.
6 . The model management device as claimed in claim 1 , wherein the processor is configured to increase the execution frequency of acquiring the data related to training of the machine learning model having the accuracy equal to or less than a predetermined lower threshold, decrease the execution frequency of acquiring the data related to training of the machine learning model having the accuracy equal to or higher than a predetermined upper threshold, and change at least one of the upper threshold and the lower threshold based on the amount of power.
7 . The model management device as claimed in claim 4 , wherein the machine learning model having the accuracy equal to or higher than the upper threshold is a machine learning model that does not output data related to human health.
8 . A model management method for managing a plurality of machine learning models using a model management device, comprising:
acquiring an accuracy of a machine learning model trained in each of a plurality of AI devices;
acquiring an amount of power that can be supplied to the plurality of AI devices; and
changing an execution frequency of acquiring data related to training of the machine learning model having the accuracy within a predetermined range based on the amount of power.
9 . A model management device for managing a plurality of machine learning models, comprising:
a communication unit capable of communicating with a plurality of AI devices, each of which continuously performs training of a machine learning model; and
a processor configured to manage training of the machine learning model in each of the plurality of AI devices, and acquire an accuracy of the machine learning model trained in each of the plurality of AI devices, wherein
the processor is configured to increase an execution frequency of acquiring data related to training of the machine learning model having the accuracy equal to or lower than a predetermined lower threshold, decrease the execution frequency of acquiring the data related to training of the machine learning model having the accuracy equal to or higher than a predetermined upper threshold, and change at least one of the lower threshold and the upper threshold so that a change amount of a power consumption amount due to the change of the execution frequency is equal to or less than a predetermined value.
10 . A model management method for managing a plurality of machine learning models using a model management device, comprising:
acquiring an accuracy of a machine learning model trained in each of a plurality of AI devices; and
increasing an execution frequency of acquiring data related to training of the machine learning model having the accuracy equal to or lower than a predetermined lower threshold, decreasing the execution frequency of acquiring the data related to training of the machine learning model having the accuracy equal to or higher than a predetermined upper threshold, and changing at least one of the lower threshold and the upper threshold so that a change amount of a power consumption amount due to the change of the execution frequency is equal to or less than a predetermined value.