Information processing method, information processing apparatus, and program
A learning procedure involves generating a non-defective product learning model by conducting machine learning using non-defective product data as teacher data, and generating a defective product learning model for each defect type by conducting machine learning for each defect type using defective product data as teacher data. A calculating procedure involves calculating the likelihood of a non-defective product from output data calculated using the non-defective product learning model to which target product data is input, and calculating the likelihood of a defective product for each defect type from output data calculated using the defective product learning model to which the target product data is input. A determining procedure involves determining that the target product data is data on a defective product having an unknown defect when the likelihood of a non-defective product and the likelihood of a defective product for each defect type satisfy a predetermined requirement.
1. An information processing method comprising:
a learning procedure involving generating a non-defective product learning model by conducting machine learning using non-defective product data as teacher data, and generating a defective product learning model for a plurality of defect types by conducting machine learning for each defect type using defective product data as teacher data;
a calculating procedure involving calculating a probability of a non-defective product from output data calculated using the non-defective product learning model to which target product data is input, and calculating a probability of a defective product for each defect type from output data calculated using the defective product learning model to which the target product data is input; and
a determining procedure involving determining that the target product data is data on a defective product having an unknown defect when the likclihoodprobability of a non-defective product and the probability of a defective product for each defect type calculated by the calculating procedure satisfy a probability-related standard.
2. The information processing method according to claim 1 , wherein the determining procedure involves making a comparison between the probability of a non-defective product and a non-defective product threshold value, and making a comparison between the probability of a defective product and a defective product threshold value for each defect type, and
the determining procedure involves determining that the target product data is data on a defective product having an unknown defect when the probabilities of a defective product for all defects are each lower than the associated defective product threshold value as a result of the comparison for each defect type and the probability of a non-defective product is lower than the non-defective product threshold value.
3. The information processing method according to claim 2 , wherein the determining procedure involves, when the probabilities of a defective product are equal to or greater than the defective product threshold values for the plurality of the defect types as a result of the comparison for each defect type, determining that one of the defect types that maximizes a difference between the probability and the defective product threshold value is the defect type for the target product data.
4. The information processing method according to claim 2 , wherein the determining procedure involves, when the probability of a defective product is equal to or greater than the associated defective product threshold value for one of the defect types as a result of the comparison for each defect type, determining that the target product data is data on a defective product having the one of the defect types.
5. The information processing method according to claim 2 , wherein the determining procedure involves determining that the target product data is non-defective product data when the probabilities of a defective product for all defects are each lower than the associated defective product threshold value as a result of the comparison for each defect type and the probability of a non-defective product is equal to or greater than the non-defective product threshold value.
6. The information processing method according to claim 1 , wherein the determining procedure involves determining whether the probability of a defective product calculated for each defect type satisfies the probability-related standard, and then determining whether the probability of a non-defective product satisfies the probability-related standard.
7. An information processing apparatus comprising:
a learner configured to generate a non-defective product learning model by conducting machine learning using non-defective product data as teacher data, and configured to generate a defective product learning model for a plurality of defect types by conducting machine learning for each defect type using defective product data as teacher data;
a calculator configured to calculate a probability of a non-defective product from output data calculated using the non-defective product learning model to which target product data is input, and configured to calculate a probability of a defective product for each defect type from output data calculated using the defective product learning model to which the target product data is input; and
a determiner configured to determine that the target product data is data on a defective product having an unknown defect when the probability of a non-defective product and the probability of a defective product for each defect type calculated by the calculator satisfy a probability-related standard.
8. A non-transitory computer readable medium that stores a program that when executed, causes a computer to function as:
a learner configured to generate a non-defective product learning model by conducting machine learning using non-defective product data as teacher data, and configured to generate a defective product learning model for a plurality of defect types by conducting machine learning for each defect type using defective product data as teacher data;
a calculator configured to calculate a probability of a non-defective product from output data calculated using the non-defective product learning model to which target product data is input, and configured to calculate a probability of a defective product for each defect type from output data calculated using the defective product learning model to which the target product data is input; and
a determiner configured to determine that the target product data is data on a defective product having an unknown defect when the probability of a non-defective product and the probability of a defective product for each defect type calculated by the calculator satisfy a probability-related standard.
9. The information processing method according to claim 1 , wherein the machine learning used for generating the non-defective product learning model and generating the defective product learning model is implemented with a variational auto-encoder learning model.
10. The information processing apparatus according to claim 7 , wherein the machine learning used for generating the non-defective product learning model and generating the defective product learning model is implemented with a variational auto-encoder learning model.
11. The non-transitory computer readable medium according to claim 8 , wherein the machine learning used for generating the non-defective product learning model and generating the defective product learning model is implemented with a variational auto-encoder learning model.