AI Paper Atlas

Research map / Detection and segmentation / CNN features

Reinforce Data, Multiply Impact: Improved Model Accuracy and Robustness with Dataset Reinforcement

Fartash Faghri, Hadi Pouransari, Sachin Mehta, Mehrdad Farajtabar, Ali Farhadi, Mohammad Rastegari, Oncel Tuzel

ICCV 2023 · Detection and segmentation › CNN features · approach: Convolution-transformer backbones · 14 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
reinforce ImageNet with ensemble-teacher outputs and augmentations (ImageNet+) instead of plain ImageNetResNet-50 · ImageNet-1k val (1000 epochs)top-1 accuracy (%): 79.6 -> 81.7Improved
use ImageNet+ instead of ImageNet for pretrainingSwinTransformer-Tiny · ImageNet-Rtop-1 accuracy (%): 37.7 -> 58.0Improved
use ImageNet+ pretraining for detection and segmentationResNet-50 · MS-COCO / ADE20KmAP / mIoU: 42.2 -> 42.5 mAP; 42.8 -> 44.2 mIoUImproved
reinforce both pretraining and fine-tuning datasetMobileNetV3-Large on Food-101 · Food-101top-1 accuracy (%): 86.1 -> 89.5Improved
use harder reinforcement (RRC+M*+R*) for light-weight CNN (ablation)MobileNetv2-0.25 · ImageNet val (150 epochs)top-1 accuracy (%): 54.3 -> 52.5Hurt
use harder reinforcement (RRC+M*+R*) for transformer (ablation)ViT-Small · ImageNet val (150 epochs)top-1 accuracy (%): RRC 77.0 -> 80.8Improved
use ensemble of IG-ResNext teachers instead of single most accurate model (ablation)dataset reinforcement teacher · KD students on ImageNetstudent accuracyImproved
increase teacher accuracy within an architecture family (ablation)KD teacher choice · KD students on ImageNetstudent accuracyMixed
swap IG-ResNext ensemble teacher for CLIP-pretrained ensemble (ablation)ImageNet+ · ImageNet val (1000 epochs), ResNet-50top-1 accuracy (%): 81.7 -> 81.1Hurt
reduce stored samples to 100 per image with top-5 sparsity (ImageNet+-Small) (ablation)ImageNet+ · ImageNet val, ResNet-50 (300 ep)top-1 accuracy (%): 80.6 -> 80.6No clear effect
use ImageNet+ rather than FKD for dataset reinforcementResNet-50 · ImageNet val (300 epochs)top-1 accuracy (%): 80.1 -> 80.6Improved

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