AI Paper Atlas

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

IDeaL: Data-Free Multi-Teacher Distillation via Improved Dead Leaves

Feyza Yavuz, Mert Bulent SARIYILDIZ, Diane Larlus

ECCV 2026 · Label-efficient and robust learning › Knowledge distillation · approach: Rethinking distillation objectives · 0 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
optimize Dead Leaves pixels with decorrelation losses (IDeaL)Dead Leaves distillation data · ImageNet val (1M samples)top-1 accuracy: 66.9 -> 78.2Improved
optimize Dead Leaves pixels with decorrelation losses (IDeaL)Dead Leaves distillation data · ImageNet val (1K samples)top-1 accuracy: 64.7 -> 74.1Improved
optimize Dead Leaves pixels with decorrelation losses (IDeaL)Dead Leaves distillation data · ADE20K (1M samples)mIoU: 29.1 -> 34.5Improved
optimize Dead Leaves pixels with decorrelation losses (IDeaL)Dead Leaves distillation data · NYUd (1M samples)RMSE: 0.632 -> 0.573Improved
optimize Dead Leaves pixels with decorrelation losses (IDeaL)Dead Leaves distillation data · Transfer (15 tasks, 1M samples)top-1 accuracy: 65.5 -> 70.6Improved
use 1K IDeaL samples instead of 1K ImageNet imagesmulti-teacher distillation, 1K samples · Transfer (15 tasks)top-1 accuracy: 65.3 -> 68.6Improved
use 1M IDeaL samples instead of 1M ImageNet imagesmulti-teacher distillation, 1M samples · ImageNet valtop-1 accuracy: 83.2 -> 78.2Hurt
replace Dead Leaves with Gaussian noise as initializationdistillation data · ImageNet val (1M samples)top-1 accuracy: 66.9 -> 22.6Hurt
replace Dead Leaves with Fractalsdistillation data · ImageNet val (1M samples)top-1 accuracy: 66.9 -> 32.7Hurt
increase Dead Leaves size from 1K to 1MDead Leaves distillation data · ImageNet valtop-1 accuracy: 64.7 -> 66.9No clear effect
optimize samples with all 4 teachers jointly instead of 1 teacher (DINO) (ablation)IDeaL · ImageNet val (10K samples)top-1 accuracy: 74.4 -> 77.0Improved
optimize jointly across 4 teachers instead of combining per-teacher sets (ablation)IDeaL 10K samples · ImageNet valtop-1 accuracy: 75.1 -> 77.0Improved

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