Beyond Spectral Clustering: Probabilistic Cuts for Differentiable Graph Partitioning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Author: Ghriss, Ayoub
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910092276793344
author Ghriss, Ayoub
author_facet Ghriss, Ayoub
contents Probabilistic relaxations of graph cuts offer a differentiable alternative to spectral clustering, enabling end-to-end and online learning without eigendecompositions, yet prior work centered on RatioCut and lacked general guarantees and principled gradients. We present a unified probabilistic framework that covers a wide class of cuts, including Normalized Cut. Our framework provides tight analytic upper bounds on expected discrete cuts via integral representations and Gauss hypergeometric functions with closed-form forward and backward. Together, these results deliver a rigorous, numerically stable foundation for scalable, differentiable graph partitioning covering a wide range of clustering and contrastive learning objectives.
format Preprint
id arxiv_https___arxiv_org_abs_2511_02272
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Beyond Spectral Clustering: Probabilistic Cuts for Differentiable Graph Partitioning
Ghriss, Ayoub
Machine Learning
Data Structures and Algorithms
Probabilistic relaxations of graph cuts offer a differentiable alternative to spectral clustering, enabling end-to-end and online learning without eigendecompositions, yet prior work centered on RatioCut and lacked general guarantees and principled gradients. We present a unified probabilistic framework that covers a wide class of cuts, including Normalized Cut. Our framework provides tight analytic upper bounds on expected discrete cuts via integral representations and Gauss hypergeometric functions with closed-form forward and backward. Together, these results deliver a rigorous, numerically stable foundation for scalable, differentiable graph partitioning covering a wide range of clustering and contrastive learning objectives.
title Beyond Spectral Clustering: Probabilistic Cuts for Differentiable Graph Partitioning
topic Machine Learning
Data Structures and Algorithms
url https://arxiv.org/abs/2511.02272