Information processing system, information processing device, information processing method, and storage medium

An information processing device comprising:

• at least one memory configured to store instructions;
• and
• at least one processor configured to execute the instructions to:
• determine a first machine learning model so as to further increase a combined loss function obtained by combining:
• a first loss function indicating a level of change in a reliability of a second image feature in a feature region of a reconstructed image of an original image, the feature region being identified by using a fourth machine learning model on the reconstructed image, from a reliability of a first image feature in a feature region of the original image identified by using the first machine learning model on the original image;
• and a second loss function indicating a level of recognition error in image recognition processing of the reconstructed image;
• determine, so as to further reduce the combined loss function, respective parameter sets for a second machine learning model used in a generation of compressed data, in which an amount of data is reduced compared to the original image, and a third machine learning model used in a generation of the reconstructed image from the compressed data;
• and
• determine a parameter set for the fourth machine learning model in common with that for the first machine learning model.

A non-transitory storage medium having stored therein a program causing a computer to perform processes as the information processing device, the processes comprising:

• determining a first machine learning model so as to further increase a combined loss function obtained by combining:
• a first loss function indicating a level of change in a reliability of a second image feature in a feature region of a reconstructed image of an original image, the feature region being identified by using a fourth machine learning model on the reconstructed image, from a reliability of a first image feature in a feature region of the original image identified by using the first machine learning model on the original image;
• and a second loss function indicating a level of recognition error in image recognition processing of the reconstructed image;
• determining, so as to further reduce the combined loss function, respective parameter sets for a second machine learning model used in a generation of compressed data, in which an amount of data is reduced compared to the original image, and a third machine learning model used in a generation of the reconstructed image from the compressed data;
• and
• determining a parameter set for the fourth machine learning model in common with that for the first machine learning model.