Learning apparatus, estimation apparatus, data generation apparatus, learning method, and computer-readable storage medium storing a learning program

A learning method in which a computer executes:

• acquiring a plurality of learning data sets that are each constituted by a combination of training data, metadata indicating an attribute regarding acquisition of the training data, and correct answer data indicating a feature included in the training data that corresponds to a correct answer to an estimation task with respect to the training data;
• and
• performing machine learning of a learning model including a first encoder for extracting domain-common features, a second encoder for extracting domain-specific features, a first metadata identifier, a second metadata identifier, and estimator,
• wherein the first encoder is configured to convert input data that is given into it to a first feature amount of the domain-common features,
• the second encoder is configured to convert the input data to a second feature amount of the domain-specific features,
• the first metadata identifier is configured to receive an output value of the first encoder, and identify an attribute regarding acquisition of the input data from the first feature amount,
• the second metadata identifier is configured to receive an output value of the second encoder, and identify an attribute regarding acquisition of the input data from the second feature amount,
• the estimator is configured to receive output values of the first encoder and the second encoder, and estimate a feature included in the input data from the first feature amount and the second feature amount,
• the performing the machine learning includes executing operations comprising:
• first training the second encoder and the second metadata identifier such that, with respect to each learning data set, an identification result obtained from the second metadata identifier by giving the training data to the second encoder matches the metadata;
• second training the first encoder, the second encoder, and the estimator such that, with respect to each learning data set, an estimation result obtained from the estimator by giving the training data to the first encoder and the second encoder matches the correct answer data;
• third training the first metadata identifier such that, with respect to each learning data set, an identification result obtained from the first metadata identifier by giving the training data to the first encoder matches the metadata;
• fourth training the first encoder such that, with respect to each learning data set, an identification result obtained from the first metadata identifier by giving the training data to the first encoder does not match the metadata;
• and
• fifth training the first encoder and the second encoder such that, with respect to each learning data set, a mutual information amount decreases between an output value obtained, as the first feature amount, from the first encoder by giving the training data to the first encoder and an output value obtained, as the second feature amount, from the second encoder by giving the training data to the second encoder, and
• performing adversarial learning on the first encoder and the first metadata identifier by repeatedly alternating between executing the third training and the fourth training until the first feature amount obtained by the first encoder does not include a component corresponding to the attribute regarding acquisition of the training data and the identification by the first metadata identifier fails.

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

• acquiring a plurality of learning data sets that are each constituted by a combination of training data, metadata indicating an attribute regarding acquisition of the training data, and correct answer data indicating a feature included in the training data that corresponds to a correct answer to an estimation task with respect to the training data;
• and
• performing machine learning of a learning model including a first encoder for extracting domain-common features, a second encoder for extracting domain-specific features, a first metadata identifier, a second metadata identifier, and estimator,
• wherein the first encoder is configured to convert input data that is given into it to a first feature amount of the domain-common features,
• the second encoder is configured to convert the input data to a second feature amount of the domain-specific features,
• the first metadata identifier is configured to receive an output value of the first encoder, and identify an attribute regarding acquisition of the input data from the first feature amount,
• the second metadata identifier is configured to receive an output value of the second encoder, and identify an attribute regarding acquisition of the input data from the second feature amount,
• the estimator is configured to receive output values of the first encoder and the second encoder, and estimate a feature included in the input data from the first feature amount and the second feature amount,
• the performing the machine learning includes executing operations comprising:
• first training the second encoder and the second metadata identifier such that, with respect to each learning data set, an identification result obtained from the second metadata identifier by giving the training data to the second encoder matches the metadata;
• second training the first encoder, the second encoder, and the estimator such that, with respect to each learning data set, an estimation result obtained from the estimator by giving the training data to the first encoder and the second encoder matches the correct answer data;
• third training the first metadata identifier such that, with respect to each learning data set, an identification result obtained from the first metadata identifier by giving the training data to the first encoder matches the metadata;
• fourth training the first encoder such that, with respect to each learning data set, an identification result obtained from the first metadata identifier by giving the training data to the first encoder does not match the metadata;
• and
• fifth training the first encoder and the second encoder such that, with respect to each learning data set, a mutual information amount decreases between an output value obtained, as the first feature amount, from the first encoder by giving the training data to the first encoder and an output value obtained, as the second feature amount, from the second encoder by giving the training data to the second encoder, and
• performing adversarial learning on the first encoder and the first metadata identifier by repeatedly alternating between executing the third training and the fourth training until the first feature amount obtained by the first encoder does not include a component corresponding to the attribute regarding acquisition of the training data and the identification by the first metadata identifier fails.