Wasserstein loss Machine Learning
noun phrase
Definition: A loss function based on the Wasserstein distance between probability distributions; in ML literature, this distance is commonly also referred to as the Earth Mover’s Distance (EMD) [Frogner et al. 2015].
Example in context: “To sum up, Wasserstein loss shows the greatest advantages and it is applied in the proposed RGAN architecture for melanoma diagnosis.” [Pérez-Perdomo, Ventura 2023]
Related terms: optimal transport loss; Wasserstein distance; Earth Mover’s Distance (EMD)