AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Multi-Prize Lottery Ticket Hypothesis: Finding Accurate Binary Neural Networks by Pruning A Randomly Weighted Network

James Diffenderfer, Bhavya Kailkhura

ICLR 2021 · Architectures and efficiency › Pruning and efficient tuning · approach: Lottery ticket discovery · 12 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 learned BatchNorm parameters to MPT-1/32 (ablation)ResNet-18 MPT · CIFAR-10Top-1 accuracy: 94.66 -> 94.8Improved
Add learned BatchNorm parameters to MPT-1/1 (ablation)VGG-Small MPT · CIFAR-10Top-1 accuracy: 88.52 -> 91.9Improved
Add learned BatchNorm parameters to MPT-1/32 (ablation)WRN-50 MPT · ImageNetTop-1 accuracy: 72.67 -> 74.03Improved
Add learned BatchNorm parameters to MPT-1/1 (ablation)WRN-34 MPT · ImageNetTop-1 accuracy: 45.06 -> 52.07Improved
Increase network depth (ablation)MPT-1/32 and MPT-1/1 · CIFAR-10Top-1 accuracyImproved
Increase network width (ablation)MPT-1/32 and MPT-1/1 · CIFAR-10Top-1 accuracyImproved
Increase pruning rate in narrower networks (ablation)MPTs · CIFAR-10Top-1 accuracyHurt
Increase pruning rate in high-width networks (ablation)MPTs · CIFAR-10Top-1 accuracyNo clear effect
Use binary weights and activations instead of binary weights with real-valued activations (ablation)MPTs · CIFAR-10 depth experimentsTop-1 accuracyMixed

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