Model generation apparatus, model generation method, computer-readable storage medium storing a model generation program, model generation system, inspection system, and monitoring system

A model generation system comprising:

• a plurality of learning apparatuses;
• and
• a model generation apparatus, wherein
• each learning apparatus comprises a first processor configured with a first program to perform operations comprising:
• collecting local learning data;
• and
• constructing a trained identification model that has acquired, by machine learning using local learning data, a capability of identifying whether given data is local learning data, and
• the model generation apparatus comprises a second processor configured to perform operations comprising:
• generating data using a generation model;
• causing, by transmitting the generated data to the learning apparatuses, the identification models of the learning apparatuses to each perform an identification on the data;
• receiving, from the learning apparatuses, results of the identification performed by the identification models of the respective learning apparatus;
• and
• training the generation model to generate data that causes an identification performance of at least one of the plurality of identification models to be degraded by performing machine learning using the received results of the identification.

A model generation method executed by a computer, comprising:

• generating data using a generation model;
• transmitting the generated data to a plurality of trained identification models that each have acquired, by machine learning using local learning data, a capability of identifying whether given data is local learning data, and causing the identification models to perform an identification on the data;
• receiving results of identification with respect to the transmitted data executed by the identification models;
• and
• training the generation model to generate data that causes an identification performance of at least one of the plurality of identification models to be degraded by performing machine learning using the received results of the identification.

A non-transitory computer-readable storage medium storing a model generation program, which when read and executed, causes a computer to perform operations comprising:

• generating data using a generation model;
• transmitting the generated data to a plurality of trained identification models that each have acquired, by machine learning using local learning data, a capability of identifying whether given data is local learning data, and causing the identification models to perform an identification on the data;
• receiving results of identification with respect to the transmitted data executed by the identification models;
• and
• training the generation model to generate data that causes an identification performance of at least one of the plurality of identification models to be degraded by performing machine learning using the received results of the identification.

An inspection system comprising:

• a plurality of inspection apparatuses;
• and
• a model generation apparatus, wherein
• each inspection apparatus comprises a first processor configured with a first program to perform operations comprising:
• collecting local learning data including product image data of a product image;
• and
• constructing a trained identification model that has acquired, by machine learning using local learning data, a capability of identifying whether given data is product image data included in the local learning data, and
• the model generation apparatus comprises a second processor configured with a second program to perform operations comprising:
• generating image data using a generation model;
• causing, by transmitting the generated image data to each inspection apparatus, the identification model of the inspection apparatus to perform an identification on the image data;
• receiving, from each inspection apparatus, a result of an identification performed by the identification model of the inspection apparatus;
• and
• training the generation model to generate image data that causes an identification performance of at least one of the plurality of identification models to be degraded by performing machine learning using the received results of the identification.

A monitoring system comprising:

• a plurality of monitoring apparatuses that each monitors a cultivation condition of a plant;
• and
• a model generation apparatus, wherein
• each monitoring apparatus comprises a first processor configured with a first program to perform operations comprising:
• collecting local learning data including condition data regarding a cultivation condition of the plant;
• and
• constructing a trained identification model that has acquired, by machine learning using local learning data, a capability of identifying whether given data is condition data included in the local learning data, and
• the model generation apparatus comprises a second processor configured with a second program to perform operations comprising:
• generating pseudo-condition data using a generation model;
• causing, by transmitting the generated pseudo-condition data to each monitoring apparatus, the identification model of the monitoring apparatus to perform an identification on the pseudo-condition data;
• receiving, from each monitoring apparatus, a result of an identification performed by the identification model of the monitoring apparatus;
• and
• training the generation model to generate pseudo-condition data that causes an identification performance of at least one of the plurality of identification models to be degraded by performing machine learning using the received results of the identification.