Explainable Multi-Task Clustering: A Mathematical Optimization Approach

E. Carrizosa, J. C. Castro Gómez, V. Guerrero

This work introduces novel approaches for Multi-Task Learning in distributed networks with locally stored data. Our goal is to partition the network into connected clusters, where nodes in a region share a single interpretable predictive model. Unlike existing methods assuming linear relationships, our framework supports a broader class of algorithms. We formalize two strategies: Model Selection (assigning a cluster a pre-trained model from a constituent node) and Model Building (training a new model using pooled cluster data). We develop a Variable Neighborhood Search (VNS) heuristic that enforces connectivity while assigning or generating models for each cluster. Furthermore, we show that specific spatial configurations can be solved exactly via Dynamic Programming. We validate our heuristic's efficiency and accuracy by comparing it against this exact method.

Keywords: Multi-Task Learning, Variable Neighborhood Search

GT AMyC III: Mathematical Optimization for Transparent Decision Making
September 4, 2026  3:30 PM
Aula 28

M. Alonso-Pena, E. Carrizosa, A. Navas-Orozco

E. Carrizosa, R. DE LEONE, M. MAGAGNINI

J. Martín-Chávez, E. Carrizosa, C. Molero-Río

C. Molero-Río, C. D'Ambrosio