AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

NSGANetV2: Evolutionary Multi-Objective Surrogate-Assisted Neural Architecture Search

Zhichao Lu, Kalyanmoy Deb, Erik Goodman, Wolfgang Banzhaf, Vishnu Naresh Boddeti

ECCV 2020 · Architectures and efficiency › Neural architecture search · approach: Bi-level differentiable NAS

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 online surrogate on near-front architectures instead of offline surrogatesMSuNAS · ImageNetarchitectures sampled / rank correlation: 350 vs 16,000 (OnceForAll) architectures; 0.9 vs 0.476 rank correlationImproved
use Adaptive Switching over MLP, CART, RBF, GP surrogates instead of a single surrogate (ablation)MSuNAS accuracy predictor · ImageNet, CIFAR-10, CIFAR-100Spearman rank-order correlationImproved
use RBF surrogate rather than MLP, CART or GP (ablation)MSuNAS · ImageNethypervolumeImproved
higher surrogate rank-order correlation (ablation)MSuNAS surrogate · ImageNethypervolumeImproved
initialize search from distribution of architectures found for related objectives (MAdds) instead of random (ablation)MSuNAS targeting accuracy and CPU latency · ImageNethypervolumeImproved
use MSuNAS instead of RL or EA methodsNAS search · CIFAR-10 / ImageNetsearch speedup: 20x faster; at least 3x faster than PNASImproved
use MSuNAS instead of NSGANet and random searchbi-objective NAS · ImageNet and CIFAR-10hypervolume per architecture evaluated: 2 - 5x fasterImproved
use NSGANetV2-s instead of MobileNetV3mobile ImageNet models · ImageNet valtop-1 accuracy (%): 75.2 -> 77.4Improved
use NSGANetV2-xl instead of EfficientNet-B1ImageNet models · ImageNet valtop-1 accuracy (%): 78.8 -> 80.4Improved
search directly on each target dataset instead of transferring architecturesnon-standard datasets · CINIC-10, STL-10, Flowers102accuracyImproved
optimize a scalarized single objective with 600M MAdds targetMSuNAS · ImageNet valtop-1 accuracy (%): 78.8 (EfficientNet-B1) -> 79.56 with 596M MAddsImproved

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