AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Learning towards Minimum Hyperspherical Energy

Weiyang Liu, Rongmei Lin, Zhen Liu, Lixin Liu, Zhiding Yu, Bo Dai, Le Song

NeurIPS 2018 · Architectures and efficiency › RNNs and normalization · approach: Neural-symbolic reasoning · 5 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
add MHE regularization (s=2)CNN-9 · CIFAR-100test error (%): 28.13 -> 27.15Improved
use half-space MHE instead of full MHE (ablation)CNN-9 · CIFAR-100test error (%): 27.15 -> 25.61Improved
Euclidean vs angular distance in MHE (ablation)CNN-9 · CIFAR-10test error (%): 6.22 -> 6.21No clear effect
add MHE to network of narrow width 16/32/64 (ablation)CNN-9 · CIFAR-100test error (%): 47.72 -> 36.84Improved
add half-space MHE to widest network 256/512/1024 (ablation)CNN-9 · CIFAR-100test error (%): 25.45 -> 21.83Improved
add MHE to 15-layer CNN (ablation)CNN-15 · CIFAR-100test error (%): N/C -> 26.9Improved
add MHE to 6-layer CNN (ablation)CNN-6 · CIFAR-100test error (%): 32.08 -> 28.16Improved
apply MHE to both hidden and output layers vs hidden only (ablation)CNN-9 MHE · CIFAR-100test error (%): 26.85 -> 26.16Improved
vary MHE hyperparameter from 10^-2 to 10^2 (ablation)CNN-9 MHE · CIFAR-100test error (%)No clear effect
add MHE to ResNet-32ResNet-32 · CIFAR-10test error (%): 5.19 -> 4.72Improved
add half-space MHE to ResNet-34ResNet-34 · ImageNet-2012 valtop-1 error (%): 30.04 -> 29.50Improved
replace orthonormal regularizer with MHEResNet-18 · ImageNet-2012 valtop-1 error (%): 33.61 -> 33.50Improved

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