AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Neural Predictor for Neural Architecture Search

Wei Wen, Hanxiao Liu, Yiran Chen, Hai Li, Gabriel Bender, Pieter-Jan Kindermans

ECCV 2020 · Architectures and efficiency › Neural architecture search · approach: Efficient NAS search strategies · 200 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
add a filtering classifier stage (two-stage predictor) (ablation)Neural Predictor · NASBench-101 cross-validationvalidation accuracy MSE: 1.95 -> 0.66Improved
replace random search with Neural PredictorNeural Predictor · NASBench-101sample efficiency: 22.83 times faster to same test accuracyImproved
replace Regularized Evolution with Neural Predictor to reach same validation accuracyNeural Predictor · NASBench-101sample efficiency: 12.40 x fasterImproved
use Regularized Evolution instead of random searchRegularized Evolution · NASBench-101 2000 modelsmean test accuracy: 93.66 -> 93.97Improved
use very small N (43 or 86) for predictor training (ablation)Neural Predictor · NASBench-101 N vs Ksearch qualityHurt
use large N=860 with small K at fixed budget (ablation)Neural Predictor · NASBench-101 N vs Ksearch qualityHurt
replace GCN with MLP predictor (ablation)Neural Predictor · NASBench-101search efficiencyHurt
replace GCN with CNN predictor (ablation)Neural Predictor · NASBench-101search efficiency: near random search performanceHurt
compare Neural Predictor against ProxylessNASNeural Predictor · ImageNet ProxylessNAS spacePareto frontier test accuracyNo clear effect
use Neural Predictor instead of random baseline with same 256 modelsNeural Predictor · ImageNet ProxylessNAS spacePareto frontier test accuracyImproved
use Neural Predictor instead of a single ProxylessNAS searchNeural Predictor · ImageNet compute costcompute: 47.5 times as much computeHurt
increase K at larger budget on NASBench-101 (ablation)Neural Predictor · NASBench-101test accuracyMixed

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