notesum.ai

Published at December 4

Theory and Fast Learned Solver for $\ell^1$-TV Regularization

math.NA

cs.NA

Released Date: December 4, 2024

Authors: Xinling Liu¹, Jianjun Wang¹, Bangti Jin²

Aff.: ¹Southwest University, Chongqing, China; ²The Chinese University of Hong Kong

Arxiv: http://arxiv.org/pdf/2412.03269v1

Refer to caption

Methods	$s_{r}=50$		$s_{r}=100$		$s_{r}=150$
Methods	$s_{g}=25$	$s_{g}=50$	$s_{g}=50$	$s_{g}=100$	$s_{g}=75$	$s_{g}=150$
	$\lambda_{1}/\lambda_{2}=1$
$\Phi_{\ell^{1}}$ [52]	400	400	661	661	870	870
$\Phi_{\rm TV}$ [17]	552	580	580	632	670	680
$\Phi(s_{r},s_{g})$	92	149	186	285	271	400
	$\lambda_{1}/\lambda_{2}=0.1$
$\Phi_{\ell^{1}}$ [52]	400	400	660	661	870	870
$\Phi_{\rm TV}$ [17]	552	580	580	632	607	680
$\Phi(s_{r},s_{g})$	508	553	556	632	599	697