notesum.ai

Published at November 3

Denoising Fisher Training For Neural Implicit Samplers

cs.LG

cs.AI

cs.CV

stat.CO

Released Date: November 3, 2024

Authors: Weijian Luo¹, Wei Deng¹

Aff.: ¹Peking University

Arxiv: http://arxiv.org/abs/2411.01453v1

Refer to caption

SGLD	DSVI
75.09% $\pm$ 0.20%	73.46% $\pm$ 4.52%
SVGD	SteinGAN
74.76% $\pm$ 0.47%	75.37% $\pm$ 0.19%
Fisher-NS	KL-IS(Ours)
76.22% $\pm$ 0.43%	75.95% $\pm$ 0.002%
DFT-NS(Ours)^∗	DFT-NS(Ours)
75.79% $\pm$ 0.55%	76.36% $\pm$ 0.009%