AI Paper Atlas

Research map / Probabilistic and scientific modeling / Differential equations and PDEs

StEik: Stabilizing the Optimization of Neural Signed Distance Functions and Finer Shape Representation

Huizong Yang, Yuxin Sun, Ganesh Sundaramoorthi, Anthony Yezzi

NeurIPS 2023 · Probabilistic and scientific modeling › Differential equations and PDEs · approach: Neural operators for general geometries · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace divergence loss with Laplacian normal loss, linear layers (ablation)DiGS (linear) · ShapeNetIoU mean: 0.9390 -> 0.9586Improved
replace divergence loss with Laplacian normal loss, linear layers (ablation)DiGS (linear) · ShapeNetsquared Chamfer mean: 1.32e-4 -> 1.71e-4Mixed
replace linear with quadratic layers, keeping divergence loss (ablation)DiGS · ShapeNetsquared Chamfer mean: 1.32e-4 -> 5.45e-5Improved
replace linear with quadratic layers, keeping divergence loss (ablation)DiGS · ShapeNetIoU mean: 0.9390 -> 0.9593Improved
replace divergence loss with Laplacian normal loss, quadratic layers (ablation)Qua+L_div · ShapeNetIoU mean: 0.9593 -> 0.9671Improved
replace divergence loss with Laplacian normal loss, quadratic layers (ablation)Qua+L_div · ShapeNetsquared Chamfer mean: 5.45e-5 -> 6.86e-5Hurt
add quadratic layers on top of Laplacian normal loss (ablation)Lin+L_L.n. · ShapeNetIoU mean: 0.9586 -> 0.9671Improved
add quadratic layers on top of Laplacian normal loss (ablation)Lin+L_L.n. · ShapeNetsquared Chamfer mean: 1.71e-4 -> 6.86e-5Improved
use StEik instead of DiGS, no normals (ablation)DiGS · SRB ground truthHausdorff distance d_H: 3.52 -> 2.800Improved
use StEik instead of DiGS, no normals (ablation)DiGS · SRB scansone-sided Chamfer d_C: 0.08 -> 0.096Hurt
use Laplacian normal loss instead of divergence loss in runtime (ablation)DiGS 5x256 · Runtime per iterationruntime: 37.86ms -> 32.52msImproved
use quadratic layers (5x128) instead of linear (5x256) (ablation)DiGS · Runtime per iterationruntime: 37.86ms -> 42.20msHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy