An Empirical Study of Self-Supervised Learning with Wasserstein Distance

An Empirical Study of Self-Supervised Learning with Wasserstein Distance

In this study, we consider the problem of self-supervised learning (SSL) utilizing the 1-Wasserstein distance on a tree structure (a.k.a., Tree-Wasserstein distance (TWD)), where TWD is defined as the L1 distance between two tree-embedded vectors. In SSL methods, the cosine similarity is often utili...

Full description

Bibliographic Details
Main Authors: Makoto Yamada, Yuki Takezawa, Guillaume Houry, Kira Michaela Düsterwald, Deborah Sulem, Han Zhao, Yao-Hung Tsai
Format: Article
Language:English
Published: MDPI AG 2024-10-01
Series:Entropy
Subjects:
optimal transport
Wasserstein distance
self-supervised learning
Online Access:https://www.mdpi.com/1099-4300/26/11/939

Internet

https://www.mdpi.com/1099-4300/26/11/939

Similar Items