AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Recall Distortion in Neural Network Pruning and the Undecayed Pruning Algorithm

Aidan Good, Jacky Lin, Xin Yu, Hannah Sieg, Mikey Fergurson, Shandian Zhe, Jerzy Wieczorek, Thiago Serra

NeurIPS 2022 · Architectures and efficiency › Pruning and efficient tuning · approach: Structured network pruning · 3 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 Undecayed Pruning instead of Magnitude PruningCIFAR-10 models · CIFAR-10, ResNet-56intensificationImproved
Use Undecayed Pruning instead of Gradient PruningCIFAR-10 models · CIFAR-10, ResNet-56intensificationImproved
Increase pruning ratioMagnitude Pruning and Undecayed Pruning · CIFAR-10, CIFAR-100, MNIST, and Fashion-MNISTintensificationHurt
Use moderate rather than high pruning ratiosPruned neural networks · Classification on balanced datasetsintensificationImproved
Use moderate pruningPruned neural networks · Classification on balanced datasetsaccuracyImproved
Use Random Pruning instead of Magnitude, Gradient, or Undecayed PruningCIFAR-10, ResNet-56 · CIFAR-10intensificationImproved
Increase model sizeMagnitude- or Undecayed-Pruned networks · CIFAR-10intensificationImproved
Increase dataset/task complexityMagnitude-Pruned networks · MNIST, Fashion-MNIST, CIFAR-10, and CIFAR-100intensificationHurt
Use Lottery Ticket Hypothesis pruning at higher pruning ratiosResNet-56 · CIFAR-10intensification: 0.986 -> 1.244Hurt
Use CHIP at a higher pruning ratioResNet-56 · CIFAR-10intensification: 0.985 -> 1.494Hurt

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