A Trainable Centrality Framework for Modern Data

TL;DR

FUSE is a neural framework that computes data centrality scores from arbitrary representations, combining global and local signals to improve robustness and efficiency in high-dimensional outlier detection and data ranking.

Contribution

Introduces FUSE, a neural centrality estimation method that integrates global and local views for flexible, efficient, and accurate data centrality measurement across diverse data types.

Findings

Recovers meaningful classical data orderings

Reveals multi-scale geometric structures

Achieves competitive outlier detection performance

Abstract

Measuring how central or typical a data point is underpins robust estimation, ranking, and outlier detection, but classical depth notions become expensive and unstable in high dimensions and are hard to extend beyond Euclidean data. We introduce Fused Unified centrality Score Estimation (FUSE), a neural centrality framework that operates on top of arbitrary representations. FUSE combines a global head, trained from pairwise distance-based comparisons to learn an anchor-free centrality score, with a local head, trained by denoising score matching to approximate a smoothed log-density potential. A single parameter between 0 and 1 interpolates between these calibrated signals, yielding depth-like centrality from different views via one forward pass. Across synthetic distributions, real images, time series, and text data, and standard outlier detection benchmarks, FUSE recovers meaningful classical ordering, reveals multi-scale geometric structures, and attains competitive performance with strong classical baselines while remaining simple and efficient.