Slide 8
Slide 8 text
Optimal transport
In recent years, optimal transport (a.k.a Earth mover’s distance,
Monge-Kantorovich problem, Wasserstein distance) has witnessed a lot of
applications in AI:
Theory (Brenier, Gangbo, Villani, Figalli, et.al.); Gradient flows
(Otto, Villani, et.al.); Proximal methods (Jordan, Kinderlehrer, Otto
et.al.);
Mean field games (Larsy, Lions et.al.);
Population games via Fokker-Planck equations (Degond et. al.
2014, Li et.al. 2016);
Image retrieval (Rubner et.al. 2000); Wasserstein of Wasserstein loss
(Dukler, Li, Lin, Guido);
Inverse problems (Stuart, Zhou, et.al., Ding, Li, Yin, Osher);
Scientific computing: (Benamou, Brenier, Carillo, Oberman, Peyre,
Solomon, Osher, Li, Yin, et. al.)
AI and Machine learning: Wasserstein Training of Boltzmann
Machines (Cuturi et.al. 2015); Learning from Wasserstein Loss
(Frogner et.al. 2015); Wasserstein GAN (Bottou et.al. 2017); see
NIPS, ICLR, ICML 2015– 2020. 8
8