arXiv：cs.LG· Haining Yu·· 3 小时前AI 评分24
通过线性规划归一化最小化 Bellman 误差
Bellman Error Minimization Via Linear Programming Normalization
AI 导读
该论文提出一种新的函数逼近方法，将深度神经网络与线性规划逼近算法结合，以降低高维动态规划与强化学习问题中的 Bellman 误差。研究以收益管理中的网络容量控制这一经典动态规划问题为动机示例，仿真结果显示所提逼近算法与基准相比具有竞争力。
正文
Abstract:This paper proposes a new functional approximation approach to reduce Bellman error in high-dimensional dynamic programming and Reinforcement Learning problems. Using a classic dynamic programming problem (network capacity control in revenue management) as the motivational example, the paper illustrates that deep neural networks and linear programming approximation algorithms can be combined to derive approximate solutions to dynamic programming problems. Simulation results show the proposed approximation algorithms achieves competitive performance when compared with benchmark.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.02730 [cs.LG] |
| (or arXiv:2610.02730v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02730 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Haining Yu [view email]
[v1]
Fri, 2 Oct 2026 03:05:07 UTC (134 KB)
来源：arXiv：cs.LG · arxiv.org