arXiv：cs.LG· Kevin Riehl, Shaimaa K. El-Baklish, Fan Wu, Anastasios Kouvelas·· 4 小时前
Kolmogorov-Arnold Networks 的样本效率研究
Sample-Efficiency of Kolmogorov-Arnold Networks
AI 导读
一项系统性计算实验显示，在 Feynman 数据集和 Gymnasium RL 基准上，Kolmogorov-Arnold 架构比 MLP 架构少用 40% 样本即可达到相近性能，训练过程中相对性能提升最高达 50%，且该增益在不同奖励噪声水平下保持稳健。该研究通过计算实验系统评估了 Kolmogorov-Arnold Networks 在强化学习中的样本效率。
正文
Abstract:Deep reinforcement learning has achieved substantial performance gains over classical control approaches. Yet, a central challenge to learning in real-world applications is acquiring costly samples. Kolmogorov-Arnold Networks are a recently proposed architecture that can learn physical relationships in control problems effectively, with significantly higher parameter efficiency and interpretability when compared to Multi-Layer-Perceptron architectures. In this work, we systematically study sample-efficiency using computational experiments, covering the Feynman dataset and the Gymnasium RL benchmark. The results show that similar performance can be achieved with 40% fewer samples using the Kolmogorov-Arnold architecture, and that relative performance improvements up to 50% occur during the training process. The observed gains are robust to varying levels of noise in rewards. These results highlight the potential of the Kolmogorov-Arnold architectures for more sample-efficient reinforcement learning. Code: this https URL
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.10627 [cs.LG] |
| (or arXiv:2610.10627v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.10627 arXiv-issued DOI via DataCite |
Submission history
From: Kevin Riehl [view email]
[v1]
Wed, 7 Oct 2026 11:17:37 UTC (357 KB)
来源：arXiv：cs.LG · arxiv.org