AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

GroSS: Group-Size Series Decomposition for Grouped Architecture Search

Henry Howard-Jenkins, Yiwen Li, Victor Adrian Prisacariu

ECCV 2020 · Architectures and efficiency › Neural architecture search · approach: Bi-level differentiable NAS · 0 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
use GroSS searched configuration instead of uniform 32s baseline at equal MACsGroSS search · CIFAR-10 4-layertrue accuracy: 83.70 -> 84.05Improved
use GroSS searched configuration instead of uniform 16s baseline, 4.25% fewer MACsGroSS search · CIFAR-10 4-layertrue accuracy: 82.94 -> 83.88Improved
use GroSS searched configuration instead of VBMF rank selectionGroSS search · CIFAR-10 4-layertrue accuracy: 82.83 -> 83.88Improved
GroSS search below uniform 32s baseline on VGG-16GroSS search · CIFAR-10 VGG-16true accuracy: 91.57 -> 91.41Hurt
GroSS search below uniform 16s baseline on VGG-16GroSS search · CIFAR-10 VGG-16true accuracy: 91.19 -> 91.31Improved
GroSS search below uniform 4s baseline on VGG-16GroSS search · ImageNet VGG-16true accuracy: 70.51 -> 70.63Improved
GroSS search below uniform 4s baseline on ResNet-18GroSS search · ImageNet ResNet-18true accuracy: 65.46 -> 65.44No clear effect
GroSS search on high-compression structure of prior workGroSS search · ImageNet VGG-16true accuracy: 66.85 -> 67.22Improved
use GroSS instead of partial fine-tuning for estimating configuration accuracyGroSS · 4-layer CIFAR-10mean top-5 average precision: 52.0 -> 71.2Improved
use GroSS instead of no fine-tuning for estimating configuration accuracyGroSS · 4-layer CIFAR-10mean top-5 average precision: 60.7 -> 71.2Improved
evaluate 45 configurations with GroSS dynamic group-size instead of separate networksGroSS · 4-layer CIFAR-10inference time: 287s -> 7sImproved

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