AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Sampling Matters in Deep Embedding Learning

Chao-Yuan Wu, R. Manmatha, Alexander J. Smola, Philipp Krahenbuhl

ICCV 2017 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 924 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
replace random sampling with semi-hard negative sampling (ablation)contrastive loss · Stanford Online Products (from scratch, ablation)Recall@1: 30.1 -> 49.4Improved
replace random sampling with distance weighted sampling (ablation)margin loss · Stanford Online Products (from scratch, ablation)Recall@1: 37.5 -> 61.7Improved
replace semi-hard sampling with distance weighted sampling (ablation)margin loss · Stanford Online Products (from scratch, ablation)Recall@1: 61.0 -> 61.7Improved
replace semi-hard sampling with distance weighted sampling (ablation)triplet l2 · Stanford Online Products (from scratch, ablation)Recall@1: 47.4 -> 54.5Improved
replace semi-hard sampling with distance weighted sampling (ablation)contrastive loss · Stanford Online Products (from scratch, ablation)Recall@1: 49.4 -> 39.2Hurt
replace triplet l2 loss with margin loss (ablation)distance weighted sampling · Stanford Online Products (from scratch, ablation)Recall@1: 54.5 -> 61.7Improved
replace contrastive loss with margin loss (ablation)semi-hard sampling · Stanford Online Products (from scratch, ablation)Recall@1: 49.4 -> 61.0Improved
pre-train on ILSVRC 2012-CLS (ablation)margin loss with distance weighted sampling · Stanford Online Products (from scratch, ablation)Recall@1: 61.7 -> 72.7Improved
learn class-specific beta instead of fixed beta=1.2 (ablation)margin loss · Stanford Online Products (pre-trained, 8K iterations)Recall@1: 71.1 -> 72.7Improved
use fixed beta=0.8 instead of 1.2 (ablation)margin loss · Stanford Online Products (pre-trained, 8K iterations)Recall@1: 71.1 -> 61.3Hurt
use example-specific beta (ablation)margin loss · Stanford Online Products (pre-trained, 8K iterations)Recall@1No clear effect
use margin loss with distance weighted sampling instead of HDCimage retrieval · CARS196 (original images)Recall@1: 73.7 -> 79.6Improved

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