Storage medium, machine learning method, and machine learning device

1. A non-transitory computer-readable storage medium storing a machine learning program that causes at least one computer to execute a process, the process comprising:

generating a plurality of machine learning models that includes a neural network by updating a parameter of a first machine learning model by using training data;

acquiring data by an image sensor;

specifying first distribution of a feature of data calculated by a second machine learning model out of the plurality of the machine learning models according to an input of data to the second machine learning model, the feature being a value output by a neuron of an output layer of the neural network;

determining whether or not output accuracy of the second machine learning model decreases based on the first distribution;

when the determining determines that the output accuracy does not decrease, generating a third machine learning model to use for a machine learning by updating a parameter of the second machine learning model based on a certain piece of the training data labeled based on the feature of the data; and

when the determining determines that the output accuracy decreases,

selecting, from the plurality of machine learning models, a fourth machine learning model that has second distribution of a feature of data input that is the most similar with third distribution of a feature of the training data among the plurality of the machine learning model, and

generating the third machine learning model by updating a parameter of the fourth machine learning model based on a certain piece of the training data labeled based on the feature of the data.

2. The non-transitory computer-readable storage medium according to claim 1 , wherein

the selecting includes referring to model information that includes the plurality of machine learning models, wherein

the process further comprising

adding the third machine learning model to the model information in association with fourth distribution of a feature of the certain piece.

3. The non-transitory computer-readable storage medium according to claim 1 , wherein

the first distribution includes center coordinates and a density of a cluster created by clustering the data based on an output of the second machine learning model input the data and the feature of the data, and

the selecting includes respectively comparing the center coordinates and the density of the first distribution with center coordinates and a density of the cluster associated with each of the plurality of machine learning models.

4. A machine learning method for a computer to execute a process comprising:

generating a plurality of machine learning models that includes a neural network by updating a parameter of a first machine learning model by using training data;

acquiring data by an image sensor;

specifying first distribution of a feature of data calculated by a second machine learning model out of the plurality of the machine learning models according to an input of data to the second machine learning model, the feature being a value output by a neuron of an output layer of the neural network;

determining whether or not output accuracy of the second machine learning model decreases based on the first distribution;

when the determining determines that the output accuracy does not decrease, generating a third machine learning model to use for a machine learning by updating a parameter of the second machine learning model based on a certain piece of the training data labeled based on the feature of the data; and

when the determining determines that the output accuracy decreases,

selecting, from the plurality of machine learning models, a fourth machine learning model that has second distribution of a feature of data input that is the most similar with third distribution of a feature of the training data among the plurality of the machine learning model, and

generating the third machine learning model by updating a parameter of the fourth machine learning model based on a certain piece of the training data labeled based on the feature of the data.

5. The machine learning method according to claim 4 , wherein

the selecting includes referring to model information that includes the plurality of machine learning models, wherein

the process further comprising

adding the third machine learning model to the model information in association with fourth distribution of a feature of the certain piece.

6. The machine learning method according to claim 4 , wherein

the first distribution includes center coordinates and a density of a cluster created by clustering the data based on an output of the second machine learning model input the data and the feature of the data, and

the selecting includes respectively comparing the center coordinates and the density of the first distribution with center coordinates and a density of the cluster associated with each of the plurality of machine learning models.

7. A machine learning device comprising:

an image sensor;

one or more memories; and

one or more processors coupled to the one or more memories and the one or more processors configured to

specify first distribution of a feature of data acquired by the image sensor, the feature being a value output by a neuron of an output layer of a neural network and being calculated by a second machine learning model out of the plurality of the machine learning models that includes the neural network according to an input of data to the second machine learning model;

determine whether or not output accuracy of the second machine learning model decreases based on the first distribution;

when the determining determines that the output accuracy does not decrease, generate a third machine learning model to use for a machine learning by updating a parameter of the second machine learning model based on a certain piece of the training data labeled based on the feature of the data; and

when the determining determines that the output accuracy decreases,

select, from the plurality of machine learning models, a fourth machine learning model that has second distribution of a feature of data input that is the most similar with third distribution of a feature of the training data among the plurality of the machine learning model, and

generate the third machine learning model by updating a parameter of the fourth machine learning model based on a certain piece of the training data labeled based on the feature of the data.

8. The machine learning device according to claim 7 , wherein the one or more processors is further configured to:

refer to model information that includes the plurality of machine learning models,

add the third machine learning model to the model information in association with fourth distribution of a feature of the certain piece.

9. The machine learning device according to claim 7 , wherein

the first distribution includes center coordinates and a density of a cluster created by clustering the data based on an output of the second machine learning model input the data and the feature of the data, wherein

the one or more processors is further configured to

respectively compare the center coordinates and the density of the first distribution with center coordinates and a density of the cluster associated with each of the plurality of machine learning models.