atlas

Research map / Architectures and efficiency / RNNs and normalization

Convolutional and Residual Networks Provably Contain Lottery Tickets

ICML 2022 · Architectures and efficiency › RNNs and normalization · approach: Representation geometry and collapse · 1 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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 (L+1) construction LT pruned from a random source network with ReLU activations (vs target)3-layer CNN on MNIST · MNISTtest accuracy (%): -0.08 vs targetNo clear effect
Use (L+1) construction LT for ResNet-22 with ReLUResNet-22 on CIFAR-10 · CIFAR-10test accuracy (%): -0.36 vs target (all LT); best 50% LT exceeds target slightlyNo clear effect
Use sigmoid activation in the (L+1) LT construction for ResNet-22ResNet-22 on CIFAR-10 · CIFAR-10test accuracy (%): -3.66 vs target (all LT); -0.5 for best 50%Hurt
Select the best 50% of LT repetitions rather than all (ablation)ResNet-22 LTs on CIFAR-10 · CIFAR-10test accuracy (%): +0.37 for ReLU, +0.49 for Leaky ReLU, +0.32 for Tanh, +4.2 for SigmoidHelped
Parameter count of LT vs target (ResNet-22, ReLU)ResNet-22 LT (L+1) construction · CIFAR-10nonzero prunable parameters: 1058140 vs 31069 targetHurt

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