Information processing device, information processing method, and recording medium

An information processing method comprising:

• acquiring detection data from a plurality of sensors, the plurality of sensors being associated with manufacturing devices;
• generating pseudo data of normal data and pseudo data of abnormal data, on the basis of the normal data, similar data of the normal data, the abnormal data, and similar data of the abnormal data, the normal data comprising the detection data, the similar data of the normal data and the similar data of the abnormal data having a similarity degree with past normal data and past abnormal data respectively;
• updating a first parameter that is multiplied by the pseudo data of the normal data to obtain an abnormality degree of the pseudo data of the normal data and multiplied by the pseudo data of the abnormal data to obtain an abnormality degree of the pseudo data of the abnormal data;
• and
• updating a second parameter that is multiplied by the normal data to obtain the similar data of the normal data and a third parameter that is multiplied by the abnormal data to obtain the similar data of the abnormal data,
• wherein
• the first parameter is updated such that a difference between data obtained by multiplying the pseudo data of the normal data by the first parameter and data obtained by multiplying the pseudo data of the abnormal data by the first parameter is maximized;
• and
• the second parameter and the third parameter are updated such that the difference between the data obtained by multiplying the pseudo data of the normal data by the first parameter and the data obtained by multiplying the pseudo data of the abnormal data by the first parameter is maximized.

A non-transitory computer-readable medium that stores a program that causes one or more processors to perform operations comprising:

• acquiring detection data from a plurality of sensors, the plurality of sensors being associated with manufacturing devices;
• generating pseudo data of normal data and pseudo data of abnormal data, on the basis of the normal data, similar data of the normal data, the abnormal data, and similar data of the abnormal data, the normal data comprising the detection data, the similar data of the normal data and the similar data of the abnormal data having a similarity degree with past normal data and past abnormal data respectively;
• updating a first parameter that is multiplied by the pseudo data of the normal data to obtain an abnormality degree of the pseudo data of the normal data and multiplied by the pseudo data of the abnormal data to obtain an abnormality degree of the pseudo data of the abnormal data;
• and
• updating a second parameter that is multiplied by the normal data to obtain the similar data of the normal data and a third parameter that is multiplied by the abnormal data to obtain the similar data of the abnormal data, wherein
• the first parameter is updated such that a difference between data obtained by multiplying the pseudo data of the normal data by the first parameter and data obtained by multiplying the pseudo data of the abnormal data by the first parameter is maximized;
• and
• the second parameter and the third parameter are updated such that the difference between the data obtained by multiplying the pseudo data of the normal data by the first parameter and the data obtained by multiplying the pseudo data of the abnormal data by the first parameter is maximized.