面向迁移学习的Cantor-Kantorovich度量
机器学习
2024-07-12 v1 人工智能
摘要
我们扩展了Banse等人(2023)在马尔可夫链语境下引入的Cantor-Kantorovich距离概念,以适用于马尔可夫决策过程(MDPs)。该提议度量是良定义的,并且可以在有限时域下有效地近似。随后,我们提供了数值证据,表明后者的度量在强化学习领域可以lead to有趣的应用。特别是,我们展示了它可用于预测迁移学习算法的性能。
引用
@article{arxiv.2407.08324,
title = {A Cantor-Kantorovich Metric Between Markov Decision Processes with Application to Transfer Learning},
author = {Adrien Banse and Venkatraman Renganathan and Raphaël M. Jungers},
journal= {arXiv preprint arXiv:2407.08324},
year = {2024}
}
备注
Presented at the 26th International Symposium on Mathematical Theory of Networks and Systems (Cambridge, UK)