Model generation apparatus, machine learning system, server, client, model generation method, and recording medium

A machine learning system in which a plurality of clients and a server are configured to communicate via a network, wherein

• the client includes:
• a first processor;
• and
• a first memory that includes instructions, which when executed, cause the processor to execute:
• identifying a parameter affecting misclassification of misclassification data incorrectly classified by a trained classification model, among parameters of the classification model for each type of the misclassification, based on the misclassification data incorrectly classified by the trained classification model;
• and
• generating a correction parameter obtained by correcting the parameter identified at the identifying for each type of the misclassification, and sending the correction parameter to the server, and wherein the server includes:
• a second processor;
• and
• a second memory that includes instructions, which when executed, cause the processor to execute:
• generating an integrated model including an integrated parameter obtained by integrating the correction parameters for each type of the misclassification received from the plurality of the clients.

A model generation method executed by a computer, comprising:

• identifying a parameter affecting misclassification of misclassification data incorrectly classified by a trained classification model, among parameters of the classification model for each type of the misclassification, based on the misclassification data incorrectly classified by the trained classification model;
• generating a correction parameter obtained by correcting the parameter identified at the parameter identification step for each type of the misclassification;
• and
• generating an integrated model including an integrated parameter obtained by integrating the correction parameters for each type of the misclassification.

A model generation method executed by a machine learning system in which a plurality of clients and a server are configured to communicate via a network, the model generation method comprising:

• identifying, by the client, a parameter affecting misclassification of misclassification data incorrectly classified by a trained classification model, among parameters of the classification model for each type of the misclassification, based on the misclassification data incorrectly classified by the trained classification model;
• generating, by the client, a correction parameter obtained by correcting the parameter identified at the parameter identification step for each type of the misclassification, and sending, by the client, the correction parameter to the server;
• and
• generating, by the server, an integrated model including an integrated parameter obtained by integrating the correction parameters for each type of the misclassification received from the plurality of the clients.

A non-transitory computer-readable recording medium storing a program that causes a computer to execute a process, the process comprising:

• identifying a parameter affecting misclassification of misclassification data incorrectly classified by a trained classification model, among parameters of the classification model for each type of the misclassification, based on the misclassification data incorrectly classified by the trained classification model;
• generating a correction parameter obtained by correcting the parameter identified at the parameter identification step for each type of the misclassification;
• and
• generating an integrated model including an integrated parameter obtained by integrating the correction parameters for each type of the misclassification.

A server configured to communicate with a plurality of clients via a network, the server comprising:

• a processor;
• and
• a memory that includes instructions, which when executed, cause the processor to execute:
• generating an integrated model including an integrated parameter obtained by integrating correction parameters for each type of misclassification received from the plurality of the clients, wherein
• the correction parameters identify a parameter affecting the misclassification of misclassification data incorrectly classified by a trained classification model, among parameters of the classification model for each type of the misclassification, based on the misclassification data incorrectly classified by the trained classification model, and
• the correction parameters are obtained by correcting the parameter identified for each type of the misclassification.

A client configured to communicate with a server via a network, the client comprising:

• a processor;
• and
• a memory that includes instructions, which when executed, cause the processor to execute:
• identifying a parameter affecting misclassification of misclassification data incorrectly classified by a trained classification model, among parameters of the classification model for each type of the misclassification, based on the misclassification data incorrectly classified by the trained classification model;
• and
• generating a correction parameter obtained by correcting the parameter identified at the identifying for each type of the misclassification, and sending the correction parameter to the server, wherein
• the server generates an integrated model including an integrated parameter obtained by integrating the correction parameters for each type of the misclassification received from a plurality of the clients.