AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

FairNAS: Rethinking Evaluation Fairness of Weight Sharing Neural Architecture Search

Xiangxiang Chu, Bo Zhang, Ruijun Xu

ICCV 2021 · Architectures and efficiency › Neural architecture search · approach: Bi-level differentiable NAS · 383 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 strict fairness (m models per step, accumulated gradients) instead of single-path SPOS sampling (ablation)supernet training · ImageNet search space (a), 13 modelsKendall tau: 0.6153 -> 0.9487Improved
add strict fairness instead of SPOS single-path sampling (ablation)supernet training · NAS-Bench-201Kendall tau: 0.5681 -> 0.7412Improved
add strict fairness instead of expectation-fairness uniform sampling with k=6 (ablation)supernet training · ImageNet search space (a), 13 modelsKendall tau: 0.4871 -> 0.9487Improved
add expectation fairness (SPOS) over the unfair One-Shot supernet (ablation)supernet training · ImageNet search space (a), 13 modelsKendall tau: 0.1245 -> 0.6153Improved
scale learning rate to 1/6 for uniform expectation-fairness training (ablation)Uniform (k=6) supernet · NAS-Bench-201Kendall tau: 0.3651 -> 0.4072Mixed
add strict-fairness supernet (FairNAS) instead of One-Shot or SPOS, constrained to 400M FLOPS (ablation)searched ImageNet model · ImageNet valtop-1 accuracy (%): 74.6 -> 75.3Improved
add SE, Swish and AutoAugment to FairNAS-AFairNAS-A · ImageNet valtop-1 accuracy (%): 75.3 -> 77.5Improved
use FairNAS instead of GDAS on NAS-Bench-201NAS-Bench-201 search · ImageNet16-120 testaccuracy (%): 41.71 -> 42.19Improved
use FairNAS instead of GDAS on NAS-Bench-201NAS-Bench-201 search · CIFAR-10 testaccuracy (%): 93.61 -> 93.23Hurt
use FairNAS backbone instead of MixNet-M in RetinaNetRetinaNet · COCO val2017AP (%): 31.3 -> 32.4Improved
use FairNAS-A instead of MobileNetV2 in DeepLabv3+DeepLabv3+ · PASCAL VOC 2012mIOU: 75.32 -> 78.54Improved
use the multi-objective EA search (NSGA-II with reinforced mutation, PPO) instead of random search or MnasNet RL (ablation)stage 2 search · ImageNet search space (a)Pareto front qualityImproved

Later papers that compared with this method

3 later accepted papers report FairNAS as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.

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