MCD: Marginal Contrastive Discrimination for conditional density estimation

TL;DR

MCD introduces a novel approach to conditional density estimation by factorizing the density into two parts and using classification techniques, outperforming existing methods on various datasets.

Contribution

The paper proposes Marginal Contrastive Discrimination, a new method that reformulates conditional density estimation into a classification problem, enabling the use of neural networks.

Findings

Significantly outperforms existing methods on multiple datasets

Leverages supervised learning techniques like neural networks

Effective across various density models and regression datasets

Abstract

We consider the problem of conditional density estimation, which is a major topic of interest in the fields of statistical and machine learning. Our method, called Marginal Contrastive Discrimination, MCD, reformulates the conditional density function into two factors, the marginal density function of the target variable and a ratio of density functions which can be estimated through binary classification. Like noise-contrastive methods, MCD can leverage state-of-the-art supervised learning techniques to perform conditional density estimation, including neural networks. Our benchmark reveals that our method significantly outperforms in practice existing methods on most density models and regression datasets.