Optimal Transport
Optimal transport (OT) is a mathematical framework for efficiently moving probability distributions from one configuration to another while minimizing a cost function, often visualized as the "earth mover's distance." Current research focuses on developing robust and scalable OT algorithms, particularly for high-dimensional data and applications involving noisy or unbalanced distributions, often employing neural networks and Sinkhorn iterations. These advancements are significantly impacting diverse fields, including machine learning (e.g., generative modeling, domain adaptation), image processing, and even economic modeling, by providing powerful tools for data analysis, representation learning, and fair data manipulation.
Papers
How optimal transport can tackle gender biases in multi-class neural-network classifiers for job recommendations?
Fanny Jourdan, Titon Tshiongo Kaninku, Nicholas Asher, Jean-Michel Loubes, Laurent Risser
TOT: Topology-Aware Optimal Transport For Multimodal Hate Detection
Linhao Zhang, Li Jin, Xian Sun, Guangluan Xu, Zequn Zhang, Xiaoyu Li, Nayu Liu, Qing Liu, Shiyao Yan
OTRE: Where Optimal Transport Guided Unpaired Image-to-Image Translation Meets Regularization by Enhancing
Wenhui Zhu, Peijie Qiu, Oana M. Dumitrascu, Jacob M. Sobczak, Mohammad Farazi, Zhangsihao Yang, Keshav Nandakumar, Yalin Wang
Optimal Transport Guided Unsupervised Learning for Enhancing low-quality Retinal Images
Wenhui Zhu, Peijie Qiu, Mohammad Farazi, Keshav Nandakumar, Oana M. Dumitrascu, Yalin Wang