AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Towards Accurate and Compact Architectures via Neural Architecture Transformer

Yong Fang Guo, Yin Zheng, Mingkui Tan, Qi Chen, Zhipeng Li, Jian Chen, Peilin Zhao, Junzhou Huang

TPAMI 2021 · Architectures and efficiency › Neural architecture search · approach: Bi-level differentiable NAS · 106 citations (OpenAlex)

Read the paper

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
enlarge transition space from NAT to NAT++ (two-level type and kernel transitions)NAT on VGG16 · CIFAR-10accuracy (%): 96.04 -> 96.16Improved
apply NAT++ to ResNet56ResNet56 · CIFAR-10accuracy (%): 93.21 -> 95.47Improved
apply NAT++ to DARTS architectureDARTS · ImageNettop-1 accuracy (%): 73.1 -> 73.9Improved
apply NAT++ to MobileNetV2 (already efficient ops)MobileNetV2 · ImageNettop-1 accuracy (%): 72.0 -> 72.5No clear effect
enlarge transition space from NAT to NAT++ on random architectures (ablation)NAT · CIFAR-10 (20 random architectures)test accuracy (%): 96.56 -> 96.79Improved
enlarge transition space from NAT to NAT++ on random architectures (ablation)NAT · CIFAR-10 (20 random architectures)#Params (M): 4.67 -> 3.66Improved
increase m from 1 to 10 in NAT (ablation)NAT · CIFAR-10 (20 random architectures)accuracy (%): 96.56 -> 96.61No clear effect
increase m from 1 to 30 in NAT (ablation)NAT · CIFAR-10 (20 random architectures)search cost (GPU hour): 5.3 -> 122.7Hurt
increase n from 1 to 30 in NAT++ (ablation)NAT++ · CIFAR-10 (20 random architectures)accuracy (%): 96.79 -> 96.84No clear effect
increase n from 1 to 30 in NAT++ (ablation)NAT++ · CIFAR-10 (20 random architectures)search cost (GPU hour): 5.7 -> 86.7Hurt
use 1-layer GCN instead of 2-layer GCN (ablation)NAT/NAT++ controller · CIFAR-10 (20 random architectures)accuracyHurt
use 5 or 10 GCN layers instead of 2 (ablation)NAT/NAT++ controller · CIFAR-10 (20 random architectures)accuracyHurt

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