AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Neuron-level Structured Pruning using Polarization Regularizer

Tao Zhuang, Zhixuan Zhang, Yuheng Huang, Xiaoyi Zeng, Kai Shuang, Xiang Li

NeurIPS 2020 · Architectures and efficiency › Pruning and efficient tuning · approach: Structured network pruning · 64 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 L1 regularizer on BN scaling factors with polarization regularizernetwork slimming ResNet-56 · CIFAR-10pruned accuracy (%): 93.27 -> 93.83Improved
replace L1 regularizer with polarization regularizernetwork slimming ResNet-56 · CIFAR-100pruned accuracy (%): 71.40 -> 72.46Improved
replace L1 regularizer with polarization regularizernetwork slimming VGG-16 · CIFAR-100pruned accuracy (%): 74.20 -> 74.25No clear effect
replace L1 regularizer with polarization regularizernetwork slimming ResNet-50 · ImageNetpruned top-1 accuracy (%): 74.88 -> 75.63Improved
replace L1 regularizer with polarization regularizer at ~70% FLOPs reductionnetwork slimming ResNet-50 · ImageNetpruned top-1 accuracy (%): 70.61 -> 74.15Improved
replace L1 regularizer with polarization regularizer at ~57% FLOPs reductionnetwork slimming MobileNet v2 · ImageNetpruned top-1 accuracy (%): 65.65 -> 67.49Improved
polarization regularizer vs uniform channel scaling at ~70% FLOPs reductionResNet-56 · CIFAR-10pruned accuracy (%): 92.25 -> 92.63Improved
polarization regularizer vs DeepHoyer at ~71% FLOPs reductionResNet-56 · CIFAR-10pruned accuracy (%): 91.26 -> 92.63Improved
increase hyper-parameter t of polarization regularizer (ablation)polarization pruning · CIFAR-10FLOPs reduction / accuracy dropMixed
increase hyper-parameter lambda (ablation)polarization pruning · CIFAR-10FLOPs reduction / accuracy dropHurt

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