Model maintenance device, pattern recognition system, model maintenance method, and computer program product

A pattern recognition system comprising:

• a memory unit that stores a model;
• a recognizing unit that estimates a label of target data for recognition using the model;
• a generating unit that generates a new model using learning data;
• an evaluating unit that
• evaluates performance of the new model using the evaluation data classified into a first group, from among evaluation data classified into a plurality of groups, and calculates a first performance evaluation value, and
• evaluates performance of the new model using the evaluation data classified into a second group, from among the evaluation data classified into a plurality of groups, and calculates a second performance evaluation value;
• and
• an updating determining unit that, based on the first performance evaluation value and the second performance evaluation value, determines whether or not existing model stored in the memory unit should be updated with the new model.

A model maintenance me or performing maintenance of a model for pattern recognition used in label estimation of target data for recognition, the model maintenance method comprising:

• generating a new model using learning data;
• evaluating that includes
• evaluating performance of the new model using evaluation data classified into a first group, from among evaluation data classified into a plurality of groups, and calculating a first performance evaluation value, and
• evaluating performance of the new model using evaluation data classified into a second group, from among the evaluation data classified into a plurality of groups, and calculating a second performance evaluation value;
• and
• determining, based on the first performance evaluation value and the second performance evaluation value, whether or not an existing model should be updated with the new model.

A computer program product having a non-transitory computer readable medium including programmed instructions, wherein the instructions, when executed by a computer, cause the computer to perform:

• generating a new model using learning data;
• evaluating that includes
• evaluating performance of the new model using evaluation data classified into a first group, from among evaluation data classified into a plurality of groups, and calculating a first performance evaluation value, and
• evaluating performance of the new model using evaluation data classified into a second group, from among the evaluation data classified into a plurality of groups, and calculating a second performance evaluation value;
• and
• determining, based on the first performance evaluation value and the second performance evaluation value, whether or not an existing model should be updated with the new model.