AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Convolutional Networks with Dense Connectivity

Gao Ming Huang, Zhuang Liu, Geoff Pleiss, Laurens J. P. van der Maaten, Kilian Q. Weinberger

TPAMI 2019 · Architectures and efficiency › CNN design · approach: Filter design and trained-filter analysis · 650 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
increase depth and growth rate (more parameters)DenseNet · C10+test error (%): 4.51 -> 3.62 -> 3.46Improved
use DenseNet instead of FractalNet with drop-path, no augmentationDenseNet (k=24, L=250) · C10test error (%): 7.33 -> 5.19Improved
use DenseNet instead of FractalNet with drop-path, no augmentationDenseNet (k=24, L=250) · C100test error (%): 28.20 -> 19.64Improved
use 250-layer instead of 100-layer DenseNetDenseNet · SVHNtest error (%): 1.76 -> 1.74No clear effect
use DenseNet (0.8M params) instead of 1001-layer pre-activation ResNet (10.2M)DenseNet k=12 L=100 · C10+test error (%): 4.62 -> 4.51No clear effect
add dense concatenation instead of summation (DenseNet vs ResNet)DenseNet · CIFAR-10+ small networksparameter efficiency: 1/3 of the parametersImproved
use smaller growth rate (24 vs 40) at 24% validation error (ablation)DenseNet · ImageNetflops: 0.88x10^10 -> 0.50x10^10Improved
vary compression factor theta 0.3, 0.5, 0.7 (ablation)DenseNet · ImageNeterror vs flopsNo clear effect
add full dense connectivity (remove 1x1 conv and compression in transitions) (ablation)DenseNet · C10+parameter and computational efficiencyHurt
restrict connections to most recent M layers (ablation)DenseNet · ImageNetvalidation errorHurt
exponentially increase growth rate per block (ablation)DenseNet · ImageNeterror vs flopsMixed
use pre-activation instead of post-activation batch normalization (ablation)DenseNet · ImageNetefficiencyImproved

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