tags | |
---|---|
|
In metric learning we map data samples onto a real-valued vector space. The task is to do it in a such a way that the relative positions of 2 data samples in the vector space reflect thier similarity. This allows to quickly compute similarity between any two data samples which is handy in retrieval.