AI Paper Atlas

Research map / Learning theory and algorithms / Kernels and regression theory

Why Linear Probing Works: Non-Vacuous Generalization Bounds via Effective Dimension

Dongxin Guo, Jikun Wu, SM Yiu

ECCV 2026 · Learning theory and algorithms › Kernels and regression theory · approach: Empirical generalization measures · 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
use spectral (covariance-aligned) prior instead of isotropic prior (ablation)PAC-Bayes bound, DINOv2-B · ImageNet-1Kbound: 2.34 -> 0.114Improved
use spectral prior instead of PCA to d_eff dims plus standard bound (ablation)PAC-Bayes bound, DINOv2-B · ImageNet-1Kbound: 0.172 -> 0.114Improved
add l2 regularisation (lambda>0) (ablation)linear probe bound, DINOv2-B · ImageNet-1Kbound: 0.205 -> 0.114Improved
use full spectrum rather than top-50 eigenvectors only (ablation)spectral bound, DINOv2-B · ImageNet-1Kbound: 0.138 -> 0.114Improved
use discriminative (DINOv2-B) rather than generative (MAE-B) pretrainingspectral bound · ImageNet-1Kbound: vacuous (>1) -> 0.114Improved
scale model from DINOv2-S to DINOv2-gspectral bound · ImageNet-1Kbound: 0.175 -> 0.112Improved
use S = gamma^2/d_eff instead of PACTran-N for rankingmodel ranking · 12 models, ImageNet-1KSpearman rho: 0.82 -> 0.91Improved
use d_eff alone (zero labels) instead of label-based metricsmodel ranking · 12 models, ImageNet-1KSpearman rho: 0.85 vs 0.82 (PACTran-N), 0.91 (S)Mixed
increase distribution shift from ImageNet source (Delta_shift 0 -> 234.1)DINOv2-B transfer bound · 7 datasetsbound: 0.114 -> 0.501Hurt
increase shift on target datasetDINOv2-B linear probe · 7 datasetsaccuracy: rho = -0.29, n.s.No clear effect

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