AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Dataset Growth

Ziheng Qin, zhaopan xu, YuKun Zhou, Kai Wang, Zangwei Zheng, Zebang Cheng, Hao Tang, Lei Shang, Baigui Sun, Radu Timofte, Xiaojiang Peng, Hongxun Yao and 1 more

ECCV 2024 · Vision-language and multimodal › Cross-modal learning · approach: Medical multimodal fusion · 2 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 MiniGPT-4 recaptioning to the pipeline (ablation)InfoGrowth on BLIP, 0.4M CC3M · MSCOCO zero-shot retrievalTR@1: 21.8 -> 34.1Improved
add gain-based sampling (ablation)InfoGrowth on BLIP, 0.4M CC3M · MSCOCO zero-shot retrievalTR@1: 21.8 -> 23.9Improved
combine recaptioning with gain-based sampling (ablation)InfoGrowth · MSCOCO zero-shot retrievalTR@1: 34.1 -> 37.1Improved
use k=4 neighbors instead of k=8 (arithmetic mean) (ablation)gain calculator · MSCOCO zero-shot retrievalTR@1: 33.6 -> 37.1Improved
use arithmetic instead of harmonic mean at k=4 (ablation)gain calculator · MSCOCO zero-shot retrievalTR@1: 35.3 -> 37.1Improved
use combined image and text gain instead of image gain only (ablation)gain calculator · MSCOCO zero-shot retrievalTR@1: 32.8 -> 37.1Improved
add alignment gain to info gain (ablation)gain calculator · MSCOCO zero-shot retrievalTR@1: 37.1 -> 33.7Hurt
process data with InfoGrowth under 25% shuffled captionsBLIP pretraining · MSCOCO zero-shot retrievalTR@1: 15.0 -> 31.2Improved
InfoGrowth data (0.4M) instead of full CC3M (2.7M)BLIP pretraining · MSCOCO zero-shot retrievalTR@1: 34.7 -> 37.1Improved
InfoGrowth data (0.40M) instead of full CC3M on Flickr30K zero-shotBLIP pretraining · Flickr30K zero-shot image-to-textR@1: 87.4 -> 82.2Hurt
choose 0.67M selected data from 2.7M candidates rather than 1.35M (ablation)InfoGrowth · MSCOCO zero-shot retrievalTR@1: 41.1 -> 40.9No clear effect
InfoGrowth selection instead of UCB dynamic pruningResNet-50 · ImageNet-1Ktop-1 accuracy (%): 75.3 -> 75.8Improved

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