AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

An Analysis of Super-Net Heuristics in Weight-Sharing NAS

Kaicheng Yu, René Ranftl, Mathieu Salzmann

TPAMI 2021 · Architectures and efficiency › Neural architecture search · approach: One-shot and few-shot supernets · 42 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
use batch statistics at test time instead of tracked running statistics (affine false) (ablation)Random-NAS super-net · NASBench-201sparse Kendall-Tau: -0.171 -> 0.783Improved
disable dynamic channeling in training and test (ablation)Random-NAS super-net · NASBench-101sparse Kendall-Tau: 0.22 -> 0.46Improved
disable dynamic channeling in training and test (ablation)Random-NAS super-net · NASBench-101stand-alone accuracy: 89.43 -> 93.65Improved
disable dynamic channeling only at test time (ablation)Random-NAS super-net · NASBench-101stand-alone accuracy: 89.43 -> 91.27Improved
use a lower learning rate 0.025 instead of 0.1 (ablation)Random-NAS super-net · NASBench-201stand-alone accuracy: 91.82 -> 92.42Improved
train for more epochs beyond 400 (ablation)Random-NAS super-net · NASBench-101, NASBench-201, DARTS-NDSsparse Kendall-Tau / final accuracyNo clear effect
use weight decay 0.0003 instead of none (ablation)Random-NAS super-net · NASBench-101sparse Kendall-Tau: -0.037 -> 0.144Mixed
reduce repeated cells from 3 to 1 (ablation)Random-NAS super-net · NASBench-201sparse Kendall-Tau: 0.751 -> 0.502Hurt
reduce batch size from 256 to 64 (ablation)Random-NAS super-net · NASBench-201sparse Kendall-Tau: 0.740 -> 0.703Hurt
reduce training portion to 0.25 (ablation)Random-NAS super-net · NASBench-201sparse Kendall-Tau: 0.751 -> 0.693Mixed
reduce initial channels from 16 to 4 (ablation)Random-NAS super-net · NASBench-201sparse Kendall-Tau: 0.740 -> 0.691No clear effect
add weight-sharing batch norm (WSBN) (ablation)Random-NAS super-net · NASBench-101sparse Kendall-Tau: 0.236 -> 0.056Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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