Search papers, labs, and topics across Lattice.
This paper introduces uncertainty-aware probabilistic constrained clustering (UPCC), which addresses the limitations of traditional pairwise constrained clustering by incorporating real-valued, entangled pairwise supervision that reflects intrinsic ambiguity and stochastic corruption. The proposed method, ECI-PP, utilizes an estimator-corrector-integrator framework to refine supervision through belief estimation and reliability-aware integration, significantly enhancing clustering performance. Experimental results demonstrate that ECI-PP consistently outperforms existing deep constrained clustering methods across various challenging benchmarks while maintaining robustness with a standard configuration.
Real-valued pairwise supervision can dramatically improve clustering performance, as shown by ECI-PP's ability to outperform state-of-the-art methods in uncertain environments.
Pairwise constrained clustering typically relies on hard must-link/cannot-link labels, whereas realistic pairwise supervision may be real-valued and entangle intrinsic ambiguity, expert judgment, and stochastic corruption. Existing deep constrained clustering (DCC) methods mainly target hard, expert-agnostic constraints, treating soft labels mostly numerically rather than semantically. We formalize this setting as uncertainty-aware probabilistic constrained clustering (UPCC), defining a canonical aleatoric target through a heterogeneous observation process and analyzing its conditional identifiability. We introduce ProbPair, an angular pairwise objective for probabilistic relations, and build ECI-PP, an estimator--corrector--integrator framework that refines imperfect supervision via belief estimation, correction, and reliability-aware integration. Across challenging probabilistic supervision settings, experiments on diverse benchmarks show that ECI-PP outperforms state-of-the-art DCC methods and remains robust with a shared default configuration.
