AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

TransBoost: Improving the Best ImageNet Performance using Deep Transduction

Omer Belhasin, Guy Bar-Shalom, Ran El-Yaniv

NeurIPS 2022 · Architectures and efficiency › Pruning and efficient tuning · approach: Low-rank compression and merging · 0 citations (OpenAlex)

Read the paper

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
replace separation loss with reciprocal (pull-together) loss (ablation)TransBoost ResNet50 · ImageNet-1k valtop-1 accuracy (%): 79.03 -> 74.64Hurt
add reciprocal pull-together loss to separation loss (ablation)TransBoost ResNet50 · ImageNet-1k valtop-1 accuracy (%): 79.03 -> 79.00No clear effect
add knowledge distillation stage (SimCLRv2 vs SimCLR)SimCLR ResNet50 transduction · ImageNet-1k valtop-1 accuracy (%): 75.95 -> 74.68Hurt
increase test set size with full training setTransBoost ResNet50 · ImageNet subsetstop-1 gain (%): +1.80 -> +2.88Improved
increase test set size with 5% training setTransBoost ResNet50 · ImageNet 5% traintop-1 gain (%): +3.42 -> +1.97Hurt
apply TransBoost to a larger ResNetTransBoost · ImageNet-1k valtop-1 gain (%): +3.60 -> +2.31Hurt
use ViT-S instead of Swin-TTransBoost · ImageNet-1k valtop-1 gain (%): +0.78 -> +2.28Improved
apply TransBoost on top of stronger recipe ResNet50-StrikesBackTransBoost · ImageNet-1k valtop-1 accuracy (%): 80.38 -> 81.15Improved
apply TransBoost to ConvNext-TTransBoost · ImageNet-1k valtop-1 accuracy (%): 82.05 -> 82.46Improved
apply TIM-GD transductive few-shot methodResNet50 · ImageNet-1k valtop-1 accuracy (%): 76.15 -> 71.51Hurt
apply Ent-min transductive few-shot methodResNet50 · ImageNet-1k valtop-1 accuracy (%): 76.15 -> 77.55Improved
apply TransBoost on CIFAR-10ResNet50 · CIFAR-10top-1 accuracy (%): 97.60 -> 97.61No clear effect

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