AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Transparent Vision: A Theory of Hierarchical Invariant Representations

Shuren Qi, Yushu Zhang, Chao Wang, Zhihua Xia, Xiaochun Cao, Fenglei Fan

ICCV 2025 · Architectures and efficiency › CNN design · approach: Biologically inspired CNN explanation · 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
replace scattering network with HIRinvariant representation · Digit benchmark (translation and rotation)F1 (%): 56.23 -> 94.98Improved
add data augmentationAlexNet · Digit benchmark (translation and rotation)F1 (%): 64.27 -> 92.93Improved
add data augmentationVGGNet · Digit benchmark (original)F1 (%): 100 -> 95.68Hurt
add data augmentationSimpleNet · Digit benchmark (original)F1 (%): 98.60 -> 48.82Hurt
use over-complete wavelet instead of complete cosine representationclassical representations · Digit benchmark (original)F1 (%): 45.43 -> 66.75Improved
swap SVM for NN classifierCosine features · AIGC forensics, 8 generators (average)F1 (%): 98.03 -> 8.14Hurt
test with random orientation and flippingVGGNet · AIGC forensics, 8 generators (average)F1 (%): 93.66 -> 89.07Hurt
test with random orientation and flippingHIR SVM · AIGC forensics, 8 generators (average)F1 (%): 98.68 -> 98.63No clear effect
test with random orientation and flippingScatterNet SVM · AIGC forensics, 8 generators (average)F1 (%): 95.77 -> 91.66Hurt
reduce training fraction from 5/5 to 1/9DenseNet · AIGC hybrid benchmarkF1 (%): 86.02 -> 76.10Hurt
reduce training fraction from 5/5 to 1/9HIR SVM · AIGC hybrid benchmarkF1 (%): 96.64 -> 94.73Hurt
use HIR instead of AIGC-specific forensic method ECCV'20AIGC forensics · AIGC hybrid benchmark (1/9)F1 (%): 79.67 -> 94.73Improved

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