AI Paper Atlas

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

Beyond Eigenfunctions: Divergence Principal Functions for Representation Learning

Ritabrata Ray, Sahil Dharod, Burak Varıcı, Nicholas Boffi, Pradeep Ravikumar

NeurIPS 2026 · Learning theory and algorithms › Kernels and regression theory · approach: RKHS kernel methods

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Generalizes spectral contrastive representation learning beyond squared-error kernel approximation to divergence-based objectives, introducing divergence principal functions as representations.

Problem

Contrastive representation learning is understood as estimating a positive-pair PMI-like kernel and factorizing it spectrally, but this analysis assumes squared-error (l2) objectives, whereas modern objectives use conditional KL, InfoNCE-style and other divergences.

Approach

Replaces the squared-error spectral geometry with a divergence-induced analogue, yielding divergence principal functions as representations that extend eigenfunction representations to non-l2 objectives.

Experimental setup

Builds on: spectral contrastive learning InfoNCE

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