Self-Certifying Classification by Linearized Deep Assignment
From MaRDI portal
Publication:6389302
arXiv2201.11162MaRDI QIDQ6389302
Author name not available (Why is that?)
Publication date: 26 January 2022
Abstract: We propose a novel class of deep stochastic predictors for classifying metric data on graphs within the PAC-Bayes risk certification paradigm. Classifiers are realized as linearly parametrized deep assignment flows with random initial conditions. Building on the recent PAC-Bayes literature and data-dependent priors, this approach enables (i) to use risk bounds as training objectives for learning posterior distributions on the hypothesis space and (ii) to compute tight out-of-sample risk certificates of randomized classifiers more efficiently than related work. Comparison with empirical test set errors illustrates the performance and practicality of this self-certifying classification method.
Has companion code repository: https://github.com/ipa-hd/ldaf_classification
This page was built for publication: Self-Certifying Classification by Linearized Deep Assignment
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6389302)
