摘要 |
The information conversion device (10) randomly selects one or more feature vectors from feature vectors for learning to form a sample set. The information conversion device (10) selects the feature vector that is appended with a label different from a label of any of feature vectors included in the sample set and has the smallest generalized average distance from feature vectors included in the sample set as the reference vector. The information conversion device (10) learns a hyperplane using a pair of one of feature vectors appended with a label different from that of the reference vector, among the feature vectors for learning, and the reference vector. |