AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Learning Where to Look – Generative NAS Is Surprisingly Efficient

Jovita Lukasik, Steffen Jung, Margret Keuper

ECCV 2022 · Architectures and efficiency › Neural architecture search · approach: Efficient NAS search strategies · 16 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 latent space optimization with weighted retraining (ablation)AG-Net · NAS-Bench-101val accuracy: 94.38 -> 94.90Improved
add latent space optimization (ablation)AG-Net · NAS-Bench-101test accuracy: 93.78 -> 94.18Improved
add latent space optimization (ablation)AG-Net · NAS-Bench-201 CIFAR-100test accuracy: 71.83 -> 73.51Improved
add latent space optimization (ablation)AG-Net · NAS-Bench-201 ImageNet16-120test accuracy: 45.04 -> 46.43Improved
add backpropagation through the surrogate into the generator (ablation)AG-Net · NAS-Bench-101val accuracy: 94.71 -> 94.90Improved
add backpropagation through the surrogate into the generator (ablation)AG-Net · NAS-Bench-201 CIFAR-100test accuracy: 73.22 -> 73.51Improved
add backpropagation through the surrogate (ablation)AG-Net · NAS-Bench-201 CIFAR-10val accuracy: 91.60 -> 91.60No clear effect
replace XGB predictor with end-to-end learnable predictor (ablation)AG-Net · NAS-Bench-101val accuracy: 94.61 -> 94.90Improved
add ranking loss to XGB predictorAG-Net · NAS-Bench-101val accuracy: 94.61 -> 94.60No clear effect
replace XGB with AG-Net predictorAG-Net · NAS-Bench-NLPvalidation perplexity: 95.95 -> 95.86Hurt
increase query budget from 192 to 400AG-Net · NAS-Bench-201 CIFAR-10val accuracy: 91.60 -> 91.61Improved
increase queries from 192 to 400AG-Net · NAS-Bench-201 ImageNet16-120val accuracy: 46.64 -> 46.73Improved

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