AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Learning to Weight Samples for Dynamic Early-Exiting Networks

Yizeng Han, Yifan Pu, Zihang Lai, Chaofei Wang, Shiji Song, Junfeng Cao, Wenhui Huang, Chao Deng, Gao Huang

ECCV 2022 · Architectures and efficiency › RNNs and normalization · approach: Efficient network compression and training · 51 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 sample weighting via weight prediction network7-exit MSDNet · CIFAR-100 anytime prediction, exit 1accuracy: 61.07 -> 62.47Improved
add learned sample weighting7-exit MSDNet · CIFAR-100 anytime prediction, exit 7accuracy: 74.50 -> 75.81Improved
add learned sample weighting6-exit RANet · CIFAR-100 anytime prediction, exit 1accuracy: 63.41 -> 65.33Improved
add learned sample weighting6-exit RANet · CIFAR-100 anytime prediction, exit 6accuracy: 71.78 -> 72.45Improved
add sample weighting on top of IMTAMSDNet + IMTA · CIFAR-100 anytime prediction, exit 7accuracy: 75.19 -> 76.31Improved
add sample weighting to 5-exit MSDNet5-exit MSDNet · CIFAR-100 dynamic inference at ~15M Mul-Addstop-1 accuracy: increased by about 1.3%Improved
add sample weighting to RANetRANet · CIFAR-100 at ~15M Mul-Addstop-1 accuracy: ~1.4%Improved
add sample weighting to MSDNetMSDNet · ImageNet latencycomputation time to reach 75.4% accuracy: 70% of original time (1.4x speed-up)Improved
replace learned weighting by vanilla selection (loss only on allocated training samples) (ablation)5-exit MSDNet · CIFAR-100accuracy: drastic drop in accuracyHurt
replace learned weights with hand-designed fixed weights 0.6 to 1.4 (ablation)5-exit MSDNet · CIFAR-100accuracy: gain limitedNo clear effect
freeze WPN instead of joint optimization (ablation)5-exit MSDNet · CIFAR-100accuracy: joint optimization outperforms frozen WPNHurt
use meta objective on samples that actually exit instead of whole meta set (ablation)5-exit MSDNet · CIFAR-100accuracy: higher accuracy at larger budgetsImproved

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