AI Paper Atlas

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

Contrastive Bayesian Analysis for Deep Metric Learning

Shichao Kan, Zhiquan He, Yigang Cen, Yang Li, Vladimir M. Mladenovic, Zhihai He

TPAMI 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 31 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
use log averaging of conditional probabilities instead of identity (ablation)CBML · CUBRecall@1: 69.2 -> 69.9Improved
use log averaging instead of sqrt (ablation)CBML · SOPRecall@1: 78.6 -> 79.9Improved
use log averaging instead of sqrt (ablation)CBML · In-ShopRecall@1: 89.7 -> 92.3Improved
use log averaging instead of sqrt (ablation)CBML · CUBRecall@1: 70.0 -> 69.9Hurt
add metric variance constraint (lambda 0.0 -> 1.0) (ablation)CBML · CUBRecall@1: 63.8 -> 69.9Improved
increase embedding size from 64 to 512 (ablation)CBML · CUBRecall@1: 64.3 -> 69.9Improved
increase embedding size from 512 to 1024 (ablation)CBML · CUBRecall@1: 69.9 -> 69.0Hurt
increase embedding size from 512 to 1024 (ablation)CBML · In-ShopRecall@1: 92.3 -> 92.5No clear effect
use CBML instead of Margin-style best prior (MS) at 64-dCBML · CUBRecall@1: 57.4 -> 64.3Improved
use CBML instead of MS at 64-dCBML · CarsRecall@1: 77.3 -> 83.2Improved
use CBML instead of A-BIER with GoogLeNetCBML · In-ShopRecall@1: 83.1 -> 88.7Improved
use CBML instead of LoOp+MS with GoogLeNetCBML · SOPRecall@1: 76.6 -> 74.8Hurt

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