Pairwise ranking loss use two equations for negative samples and positive samples. For postive samples, we hope the relative distance is as small as possible. For negative samples, we hope the relative distance is larger than a margin value at least.

University of Michigan - Ann Arbor

Google

Ranking loss focuses more on relative distance instead of measuring the direct difference between the output with the expectation. 

Ranking Loss

Goodfellow, I., Bengio, Y., & Courville, A. (2016). $\mathit{Deep \ Learning.}$ MIT Press. Retrieved from [www.deeplearningbook.org](https://www.deeplearningbook.org) 

Deep Learning

Reference of Foundations of Large Language Models Course

Pairwise Ranking Loss

Every sample used for training using a triplet ranking loss is made up of three components, an anchor, a positive sample and a negative sample. Triplet loss aims at making the relative distance between the anchor and the positive sample is margin value larger than the relative distance between the anchor and the negative sample.

Learn Before

Related