notesum.ai

Published at November 22

Self-Supervised Learning for Ordered Three-Dimensional Structures

Released Date: November 22, 2024

Authors: Matthew Spellings¹, Maya Martirossyan², Julia Dshemuchadse²

Aff.: ¹Vector Institute; ²Cornell University

Arxiv: http://arxiv.org/abs/2411.14680v1

Refer to caption

	Frame	Noisy	Autoencoder	Denoising	Nearest	Shift	$Q_{\ell}$	$\|\Psi\|$	$\frac{\|r_{n}\|}{\|r_{0}\|}$
$L_{A}$ / $L_{A}$	0.512	0.509	0.512	0.505	0.513	0.509	0.504	0.504	0.509
$G_{A}$ / $L_{A}$	1.000	1.000	1.000	0.918	0.999	1.000	0.999	0.998	0.972
$L_{A}$ / $S_{A}$	0.726	0.815	0.570	0.579	0.537	0.584	0.903	0.955	0.752
$G_{A}$ / $S_{A}$	0.957	0.958	0.955	0.805	0.969	0.965	0.949	0.897	0.909
$L_{A}$ / $L_{B}$	0.962	0.530	0.859	0.814	0.690	0.622	0.506	0.505	0.530
$S_{A}$ / $S_{B}$	1.000	0.845	0.895	0.828	0.644	0.704	0.750	0.788	0.582