AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Sparse Weight Activation Training

Md Aamir Raihan, Tor Aamodt

NeurIPS 2020 · Architectures and efficiency › RNNs and normalization · approach: Efficient network compression and training · 10 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
SWAT-U instead of other sparse training methods at 80% weight sparsityResNet-50 · ImageNet valtop-1 accuracy (%): 75.2 vs 74.9 (SNFS), 74.6 (RigL)Improved
sparsify weights and activations in both forward and backward pass (SWAT-U at 80%)ResNet-50 · ImageNet valtop-1 accuracy change: -1.6 accuracy change at 76.1% FLOPs reductionHurt
use ERK layerwise sparsity instead of uniform (ablation)SWAT ResNet-50 · ImageNet val 80% sparsitytop-1 accuracy (%): 75.2 -> 76.0Improved
structured SWAT channel pruning at 70% sparsityResNet-18, DenseNet-121 · CIFAR-10top-1 accuracy: +0.06 and +0.30 accuracy changeImproved
structured SWAT channel pruning at 70% sparsityResNet-50 · ImageNet valtop-1 accuracy (%): 76.80 -> 75.67Hurt
mask gradient updates for non-active weights (ablation)SWAT ResNet-18 · CIFAR-100validation accuracyHurt
freeze sparse topology early in training (ablation)SWAT ResNet-18 · CIFAR-100validation accuracyHurt
Top-K selection instead of random selection of components (ablation)SAW ResNet-18 · CIFAR-100accuracyImproved
constrain Top-K per channel (TOPK-HW) instead of whole tensor (ablation)SAW ResNet-18 · CIFAR-100accuracyHurt
compute Top-K threshold periodically (every 100 iterations) (ablation)SWAT ResNet-18 · CIFAR-100, 90% target sparsitysparsity stabilityNo clear effect
sparsify output activation gradients (meProp) instead of weights and activations (SAW) (ablation)ResNet-18 · ImageNetconvergence/accuracyHurt

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