Implicit Modeling -- A Generalization of Discriminative and Generative Approaches

TL;DR

This paper introduces an implicit modeling approach that unifies discriminative and generative models by specifying only conditional distributions, enhancing flexibility and generalization in classification tasks.

Contribution

It presents a novel implicit modeling framework that combines the strengths of discriminative and generative models, allowing for more powerful and generalizable classifiers.

Findings

Effective on artificial classification data

Advantages demonstrated in semantic image segmentation

Improved generalization capabilities

Abstract

We propose a new modeling approach that is a generalization of generative and discriminative models. The core idea is to use an implicit parameterization of a joint probability distribution by specifying only the conditional distributions. The proposed scheme combines the advantages of both worlds -- it can use powerful complex discriminative models as its parts, having at the same time better generalization capabilities. We thoroughly evaluate the proposed method for a simple classification task with artificial data and illustrate its advantages for real-word scenarios on a semantic image segmentation problem.