1. Self-Certifying Classification by Linearized Deep Assignment
- Author
-
Boll, Bastian, Zeilmann, Alexander, Petra, Stefania, and Schnörr, Christoph
- Subjects
Statistics - Machine Learning ,Computer Science - Machine Learning ,Mathematics - Optimization and Control - 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.
- Published
- 2022