AI Paper Atlas

Research map / Label-efficient and robust learning / Knowledge distillation

Simultaneous Similarity-based Self-Distillation for Deep Metric Learning

Karsten Roth, Timo Milbich, Bjorn Ommer, Joseph Paul Cohen, Marzyeh Ghassemi

ICML 2021 · Label-efficient and robust learning › Knowledge distillation · approach: Patient consistent distillation · 4 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 S2SD to DML objectivesMargin, R-Margin, and Multisimilarity · CUB200-2011, CARS196, and SOPRecall@1: up to $7\%$Improved
Add dual similarity self-distillation from one higher-dimensional branch (ablation)Margin DML · CUB200-2011Recall@1: $63.09\pm0.46$ -> $65.11\pm0.18$Improved
Add multiscale higher-dimensional target branches instead of a single target branch (ablation)Margin DML · CUB200-2011Recall@1: $65.11\pm0.18$ -> $66.13\pm0.34$Improved
Add feature-space distillation to MSD (ablation)Margin DML · CUB200-2011Recall@1: $66.13\pm0.34$ -> $67.58\pm0.32$Improved
Add multiple pooling operations in auxiliary branches (ablation)Margin + MSDF · SOPRecall@1: $79.63\pm0.14$ -> $78.82\pm0.09$Hurt
Distill each target space directly to the base space rather than using nested or chained distillation (ablation)Multiscale self-distillation · DML benchmarksRecall@1Improved
Use similarity-vector KL distillation rather than full-matrix/mean KL or cosine/euclidean matching (ablation)S2SD · DML benchmarksRecall@1Improved
Use a two-layer MLP for auxiliary branches rather than a linear mapping (ablation)S2SD · DML benchmarksRecall@1Improved
Increase auxiliary branch depth beyond two-layer MLPs (ablation)S2SD · DML benchmarksRecall@1Hurt
Train auxiliary spaces jointly without similarity distillation (ablation)S2SD · CUB200-2011Recall@1: $62.70\pm0.53$ -> $62.93\pm0.22$No clear effect
Detach the low-dimensional reference embedding while training in high dimension (ablation)S2SD · CUB200-2011Recall@1: $62.70\pm0.53$ -> $63.08\pm0.42$No clear effect
Replace simultaneous distillation with standard two-stage distillation (ablation)DML student, d=128 · CUB200-2011Recall@1: $65.57\pm0.26$ -> $63.89\pm0.14$Hurt

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