notesum.ai

Published at December 3

cs.IT

eess.SP

math.IT

Released Date: December 3, 2024

Authors: Ziheng Liu¹, Jiayi Zhang¹, Yiyang Zhu¹, Enyu Shi¹, Bo Ai¹

Aff.: ¹Beijing Jiaotong University

Parameters	Size
1st hidden layer (MADDPG)	128, Leaky Relu (0.01)
2nd hidden layer (MADDPG)	64, Leaky Relu (0.01)
Hidden layer (DHPN and GNN)	128 and 64 Relu
Hidden layer (RNN)	256, Relu
Number of heads (DHPN and GNN)	8 and 4
Discounted factor $\gamma$ , $\gamma_{\mathrm{a}}$ , and $\gamma_{\mathrm{ex}}$	0.99
Experience extraction pool $\mathcal{P}_{s}$	1024
Maximum gradient clipping value $\xi$	0.5
Soft update rate and learning rate	0.01