Autonomy-Aware Clustering: When Local Decisions Supersede Global Prescriptions

ICLR 2026 Conference Submission15625 Authors

Keywords: Clustering, Local Autonomy, Facility Location, Reinforcement Learning, Deep Learning

TL;DR: We introduce autonomy-aware clustering, a reinforcement-based method that models latent data autonomy—where points may override prescribed assignments—altering cluster structure.

Abstract: Clustering arises in a wide range of problem formulations, yet most existing approaches assume that the entities under clustering are passive and strictly conform to their assigned groups. In reality, entities often exhibit local autonomy, overriding prescribed associations in ways not fully captured by feature representations. Such autonomy can substantially reshape clustering outcomes—altering cluster compositions, geometry, and cardinality—with significant downstream effects on inference and decision-making. We introduce autonomy-aware clustering, a reinforcement (RL) learning framework that learns and accounts for the influence of local autonomy without requiring prior knowledge of its form. Our approach integrates RL with a deterministic annealing (DA) procedure, where, to determine underlying clusters, DA naturally promotes exploration in early stages of annealing and transitions to exploitation later. We also show that the annealing procedure exhibits phase transitions that enable design of efficient annealing schedules. To further enhance adaptability, we propose the Adaptive Distance Estimation Network (ADEN), a transformer-based attention model that learns dependencies between entities and cluster representatives within the RL loop, accommodates variable-sized inputs and outputs, and enables knowledge transfer across diverse problem instances. Empirical results show that our framework closely aligns with underlying data dynamics: even without explicit autonomy models, it achieves solutions close to the ground truth (gap $\sim$3–4\%), whereas ignoring autonomy leads to substantially larger gaps ($\sim$35–40\%).

Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning

Submission Number: 15625