AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Beyond Filter Pruning: Top-K Spatial Selection for Efficient Neural Networks

Sarthak Ketanbhai Modi, Hans Soegeng, Thomas Peyrin

ECCV 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Structured network pruning · 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
prune spatial positions within kernels (TKP, K_keep=1) instead of filter pruning (DepGraph)ResNet56 · CIFAR-10top-1 accuracy: 88.26 -> 91.80Improved
TKP instead of ETPVGG-19 · CIFAR-100top-1 accuracy: 71.30 -> 72.77Improved
TKP on ResNet50 at 3.51x FLOP reductionResNet50 · ImageNet-1ktop-1 accuracy: 76.15 -> 76.49Improved
TKP on ViT-B/16 at 1.78x FLOP reductionViT-B/16 · ImageNet-1ktop-1 accuracy: 81.07 -> 80.93No clear effect
TKP instead of MDP for measured speed-upResNet56 · CIFAR-10speed-up: 2.97x -> 3.94xImproved
TKP instead of Isomorphic pruningMobileNetV3-Large · ImageNet-1ktop-1 accuracy: 70.8 -> 72.8Improved
combine TKP with QAT instead of DepGraph with QATResNet-18 · CIFAR-100accuracy: 76.96% -> 81.54%Improved
combine TKP with QAT instead of QAT-INT8 aloneVGG-19 · CIFAR-100accuracy: 72.14% -> 71.45%Hurt
TKP instead of Wanda++ at 2:4LLaMA-2-7B · WikiText-2perplexity: 10.15 -> 9.31Improved
TKP at 50% sparsity vs denseLLaMA-2-7B · WikiText-2perplexity: 6.00 -> 5.59Improved
TKP instead of Wanda++ at 2:4BLIP-Base · Flickr30kBLEU-4: 28.93 -> 29.93Improved
add exponential regularization stage before Top-K (ablation)TKP (CNNs) · CIFAR-10/100 (ResNet-56, VGG-8, GoogleNet, DenseNet-121)accuracyImproved

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