Metric Learning

Metric learning can also be considered as similarity. Metric learning is to measure the similarity between samples, which is one of the core issues of pattern recognition.

The goal of metric learning is to minimize the distance between samples of the same type and maximize the distance between samples of different types.