AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Learning Partial Correlation Based Deep Visual Representation for Image Classification

Saimunur Rahman, Piotr Koniusz, Lei Wang, Luping Zhou, Peyman Moghadam, Changming Sun

CVPR 2023 · Architectures and efficiency › CNN design · approach: Compact CNN architectures · 16 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
replace covariance pooling with sparse inverse-covariance poolingResNet-50 visual representation · Airplaneaccuracy: 91.2 -> 92.7Improved
replace covariance pooling with sparse inverse-covariance poolingResNet-50 visual representation · Carsaccuracy: 92.1 -> 93.5Improved
replace covariance pooling with sparse inverse-covariance poolingResNet-50 visual representation · DTDaccuracy: 73.7 -> 60.7Hurt
replace covariance pooling with sparse inverse-covariance poolingResNet-50 visual representation · ImageNet100Top-1 accuracy: 71.5 -> 74.8Improved
increase matrix dimension from 256×256 to 512×512 (ablation)VGG-16 iSICE · MIT Indooraccuracy: 78.7 -> 81.1Improved
increase matrix dimension from 256×256 to 512×512 (ablation)VGG-16 iSICE · Carsaccuracy: 94 -> 94.6Improved
add learning-rate and sparsity modulators (ablation)iSICE with ResNet-50 · Airplaneaccuracy: 92.7 -> 93.4Improved
add learning-rate and sparsity modulators (ablation)iSICE with ResNet-50 · ImageNet100accuracy: 74.8 -> 76.3Improved
condition modulators on concatenated feature map and Δ1 rather than feature map alone (ablation)iSICE+MLP with ResNet-50 · ImageNet100accuracy: 76.3 -> 77.1Improved
vary sparsity constant λ across the tested range (ablation)iSICE · fine-grained datasetsaccuracy: 92.56 ±0.25No clear effect
vary learning rate η across the tested range (ablation)iSICE · Birdsaccuracy: 86.51±0.14No clear effect
vary iteration count N (ablation)iSICE · MIT Indooraccuracy: 78.28; 78.66; 78.36No clear effect

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