AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Hidden Monotonicity: Explaining Deep Neural Networks via their DC Decomposition

Jakob Paul Zimmermann, Georg Loho

CVPR 2026 · Architectures and efficiency › RNNs and normalization · approach: Representation geometry and collapse · 1 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 LayerCAM with SplitCAM on the split networkVGG16 (pretrained) · ImageNet-S-50 testPointing Game: 0.855 -> 0.938Improved
replace gamma-LRP with SplitLRPVGG16 (pretrained) · ImageNet-S-50 testPointing Game: 0.680 -> 0.871Improved
replace LayerCAM with SplitCAM on the split networkResNet18 (pretrained) · ImageNet-S-50 testPointing Game: 0.971 -> 0.972Improved
replace LayerCAM with SplitCAM on the split networkResNet18 (pretrained) · ImageNet-S-50 testSelectivity (lower is better): 3.690 -> 3.503Improved
use backward stabilization alpha=0.4 instead of alpha=0 for SplitGrad (ablation)SplitGrad (26, g), VGG16 · ImageNet-S-50 testPointing Game: 0.401 -> 0.947Improved
use alpha=0.4 instead of alpha=0.5 (scaled original gradients) for SplitGrad (ablation)SplitGrad (26, g), VGG16 · ImageNet-S-50 testPointing Game: 0.898 -> 0.947Improved
use alpha=0.4 instead of alpha=0.5 for SplitCAM (ablation)SplitCAM (26, +, g), VGG16 · ImageNet-S-50 testPointing Game: 0.868 -> 0.938Improved
use alpha=0 (no stabilization) instead of intermediate alpha (ablation)SplitGrad (26, g), VGG16 · ImageNet-S-50 testMaximum Sensitivity (lower is better): 0.616 -> 0.028Improved
very small alpha (0.3) for SplitCAM (ablation)SplitCAM (26, +, g), VGG16 · ImageNet-S-50 testPointing Game: 0.938 -> 0.613Hurt
directly train as difference of convex or monotone networks (DIC/DM) on MLP maxoutMNIST mlp maxout · MNIST testaccuracyImproved
invert the input to the h-stream (DICIN/DMIN) (ablation)DIC/DM models on MNIST · MNISTexplanation separation (qualitative)Improved
use winner-takes-all max pooling instead of the convexity-preserving identity (ablation)Split networks · ImageNet-S-50monotonicity and convexity of streamsHurt

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