AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Directional Neural Collapse Explains Few-Shot Transfer in Self-Supervised Learning

Achleshwar Luthra, Yash Salunkhe, Tomer Galanti

ICML 2026 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Invariance-based self-supervised learning · 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
track directional CDNV instead of classical CDNV during SSL pretrainingMAE, SimCLR, DINO-v2, VICReg · mini-ImageNet (train and test)directional CDNV: from roughly 2^-1 to 2^1 down to about 2^-3 to 2^-5Improved
measure classical CDNV during SSL pretrainingMAE, SimCLR, DINO-v2, VICReg · mini-ImageNetclassical CDNVNo clear effect
use the proposed finite-shot directional bound instead of the prior directional boundNCC on frozen SSL encoders · mini-ImageNet binary tasksfew-shot error certificateImproved
use the proposed bound instead of Galanti et al. 2023 boundNCC and LP few-shot error · mini-ImageNetbound tightnessImproved
train SSL longer on decision-axis varianceSSL encoders · mini-ImageNet class pairsdecision-axis varianceImproved
train SSL on synthetic multi-factor imagesSimCLR ResNet-18, VICReg ResNet-50 · synthetic color/background/shape/size labelingscosine similarity between decision axesImproved
train VICReg on real attribute dataVICReg ResNet-50 · CelebA, CUB-200cosine similarity between decision axesImproved

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