A Hybrid Tsallis-Polarization Impurity Measure for Decision Trees: Theoretical Foundations and Empirical Evaluation

This feed aggregates recent arXiv submissions (Mar 17, 2026) across statistical machine learning and related areas. Key contributions include: (1) ITC, a theoretically grounded hybrid impurity measure combining normalized Tsallis entropy with an exponential polarization term for decision trees; (2) FSPA, a projection-first algorithm for quantum PCA that avoids explicit eigenvalue estimation; (3) analysis showing standard acquisition functions suffice for asynchronous Bayesian optimization when intermediate posterior updates are considered; (4) Holographic Invariant Storage (HIS), a Vector-Sim-