Reward shaping using graph MAMBA

Jianghui Sang

doi:10.1016/j.patcog.2026.114404

Reward shaping using graph MAMBA

2026-07-08

0

PRE

AI

J

Jianghui Sang

G

Guangdi Jiang

Y

Yongli Wang

杨

杨华 (Hua Yang)

J

Jun Huang *

DOI：10.1016/j.patcog.2026.114404

Original

Original request for help

Save

Abstract

En 中文

• Propose a temporal graph MAMBA architecture for reward shaping without changing the optimal policy. • Develop a kernelized message-passing mechanism that reduces the computational complexity from quadratic to linear time. • Attain state-of-the-art performance across multiple standard benchmark platforms.

Keywords:

Markov decision process

Reinforcement learning

Reward shaping

Graph

MAMBA

Reward shaping using graph MAMBA

Abstract

Journal

Organization

Cited Papers

Citing Papers