AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Does Unsupervised Architecture Representation Learning Help Neural Architecture Search?

Shen Yan, Yu Zheng, Wei Ao, Xiao Zeng, Mi Zhang

NeurIPS 2020 · Architectures and efficiency › Neural architecture search · approach: Efficient NAS search strategies · 33 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
learn architecture embeddings unsupervised (decoupled from search) instead of supervised joint learningarch2vec RL search · NAS-Bench-101test accuracy (%): 93.74 -> 94.10Improved
learn embeddings unsupervised instead of supervisedarch2vec BO search · NAS-Bench-101test accuracy (%): 93.79 -> 94.05Improved
add KL regularization term to the autoencoder (ablation)arch2vec · NAS-Bench-101 validityvalidity (%): 30.31 -> 44.97Improved
add KL regularization term to the autoencoder (ablation)arch2vec · DARTS validityvalidity (%): 16.01 -> 33.36Improved
add KL regularization term to the autoencoder (ablation)arch2vec · NAS-Bench-201 reconstruction accuracyreconstruction accuracy: 100 -> 100No clear effect
unsupervised embeddings instead of supervised for GP performance predictionarch2vec · NAS-Bench-101RMSE: 0.038 -> 0.018Improved
use arch2vec embeddings with BO instead of discrete RENAS-Bench-201 · CIFAR-100test accuracy (%): 72.86 -> 73.37Improved
arch2vec-BO vs supervised BANANAS in DARTS spaceDARTS search space · CIFAR-10avg test error (%): 2.67 -> 2.56Improved
discrete adjacency-matrix encoding replaced by supervised continuous embeddingsRE/BOHB search · NAS-Bench-101test regretImproved

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