AI Paper Atlas

Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes

Gaussian RBF Centered Kernel Alignment (CKA) in the Large-Bandwidth Limit

Sergio A. Álvarez

TPAMI 2022 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Sparse variational Gaussian processes · 23 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
remove mean-centering of Gram matricesGaussian RBF CKA · 2x2 counterexampleconvergence to linear CKA: Gaussian CKA 1 vs linear 3/sqrt(14)Hurt
use bandwidth sigma larger than rhoGaussian RBF CKA · all data setsmean relative difference to linear CKA: less than 0.01Improved
use small bandwidth sigma <= 0.25Gaussian RBF CKA · classification data setsrelative difference to linear CKA: greater than 0.2; greater than 0.7 for dna at w>=128Improved
increase network widthMLP representations · regression data setsrepresentation eccentricity rho: cpu 10.63 -> 8.82Improved
increase network widthMLP representations on balloon · balloonrepresentation eccentricity rho: 14.62 -> 25.18Hurt
increase network widthMLP representations · classification data setsstandard error of relative CKA differenceImproved
increase rhoGaussian CKA · data setsconvergence onset log2 sigma*_0: balloon rho 16 -> onset 5; wdbc rho 11 -> 4; cpu rho 9 -> 4Hurt
raise raw data dimensionalityMLP representations · classification and regressioncorrelation with rho: -0.38 to -0.33 classification; -0.8 to -0.85 regressionImproved

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