notesum.ai

Published at October 18

Learning to refine domain knowledge for biological network inference

cs.LG

cs.AI

cs.CL

cs.CV

stat.ML

Released Date: October 18, 2024

Authors: Peiwen Li¹, Menghua Wu²

Aff.: ¹Shenzhen International Graduate School, Tsinghua University; ²Department of Computer Science, Massachusetts Institute of Technology

Arxiv: https://arxiv.org/abs/2410.14436v1

Refer to caption

Prior Knowledge	Model	Graph		Noise
Prior Knowledge		mAP $\uparrow$	SHD $\downarrow$	F1 $\uparrow$	Acc $\uparrow$
Vanilla	Dcdi-G	0.14	23	—	—
	Dcdi-Dsf	0.14	29	—	—
	BaCaDi	0.15	20	—	—
	Sea	0.20	17	—	—
Synthetic KG (10% noise)	Dcdi-G	0.14	21	0.55	0.73
	Dcdi-Dsf	0.15	18	0.53	0.75
	Ours	0.25	17	0.73	0.80
Synthetic KG (25% noise)	Dcdi-G	0.19	19	0.67	0.76
	Dcdi-Dsf	0.15	23	0.35	0.60
	Ours	0.24	17	0.78	0.80
CORUM KG (34% noise)	Dcdi-G	0.14	24	0.44	0.73
	Dcdi-Dsf	0.15	20	0.67	0.82
	Ours	0.21	16	0.70	0.78