AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Expressive yet Efficient Feature Expansion with Adaptive Cross-Hadamard Products

Xuyang Zhang, Xi Zhang, Liang Chen, Hao Shi, Qingshan Guo

ICLR 2026 · Architectures and efficiency › Neural architecture search · approach: AutoML pipeline search · 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
use softsign instead of sigmoid in dynamic normalization (ablation)ACH small model · CIFAR-100top-1 accuracy (%): 73.14 -> 73.57Improved
use softsign instead of algebraic sigmoid (ablation)ACH small model · CIFAR-100top-1 accuracy (%): 72.80 -> 73.57Improved
replace IB9 and IB10 with ACH (ablation)MobileNetV3-S · CIFAR-100top-1 accuracy (%): 70.01 -> 71.58Improved
replace only IB0 with ACH (ablation)MobileNetV3-S · CIFAR-100top-1 accuracy (%): 70.01 -> 69.74Hurt
replace only IB1 with ACH (ablation)MobileNetV3-S · CIFAR-100top-1 accuracy (%): 70.01 -> 69.74Hurt
replace last two layers with ACH (ablation)MobileNetV3-S · CIFAR-100top-1 accuracy (%): 70.01 -> 71.58Improved
replace last two layers with ACH (ablation)ShuffleNetV2-1.0 · CIFAR-100top-1 accuracy (%): 65.89 -> 71.68Improved
replace last two layers with ACH (ablation)StarNet-S1 · CIFAR-100top-1 accuracy (%): 71.84 -> 72.07Improved
replace last two layers with ACH (ablation)MobileNetV4-S · CIFAR-100top-1 accuracy (%): 73.15 -> 72.19Hurt
replace ACH modules in MobileViT-small (ablation)MobileViT-small · ImageNet-1ktop-1 accuracy (%): 71.70 -> 72.42Improved
replace ACH modules in MobileViT-xx-small (ablation)MobileViT-xx-small · ImageNet-1ktop-1 accuracy (%): 67.62 -> 67.42No clear effect
use parity-balanced indexing instead of native ACHGPU ACH kernel · GPU latencylatency (ms): 12.40 -> 7.13Improved

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