| 摘要 |
PROBLEM TO BE SOLVED: To provide a technique that efficiently creates a plurality of predomains from a plurality of data collected in order to create predomains.SOLUTION: In a clustering device 10, a feature extraction unit 11 extracts a feature from each of a plurality of transfer candidate data 41 used in machine learning for which transfer learning is introduced, and creates a plurality of transfer candidate feature data 42. A classification unit 12 classifies the transfer candidate feature data 42 into a plurality of groups including a first group on the basis of a feature amount that each of the plurality of transfer candidate feature data 42 possesses. A predomain determination unit 14 determines that the first group is a predomain when the number of transfer candidate feature data 42 classified into the first group is less than or equal to a prescribed classification continuance reference value, and determines that the transfer candidate feature data 42 classified into the first group be further classified when the number of transfer candidate feature data 42 is larger than the classification continuance reference value.SELECTED DRAWING: Figure 2 |
