Contrastive Predictive Coding Done Right for Mutual Information Estimation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ryu, J. Jon, Yeddanapudi, Pavan, Xu, Xiangxiang, Wornell, Gregory W.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915586599026688
author Ryu, J. Jon
Yeddanapudi, Pavan
Xu, Xiangxiang
Wornell, Gregory W.
author_facet Ryu, J. Jon
Yeddanapudi, Pavan
Xu, Xiangxiang
Wornell, Gregory W.
contents The InfoNCE objective, originally introduced for contrastive representation learning, has become a popular choice for mutual information (MI) estimation, despite its indirect connection to MI. In this paper, we demonstrate why InfoNCE should not be regarded as a valid MI estimator, and we introduce a simple modification, which we refer to as InfoNCE-anchor, for accurate MI estimation. Our modification introduces an auxiliary anchor class, enabling consistent density ratio estimation and yielding a plug-in MI estimator with significantly reduced bias. Beyond this, we generalize our framework using proper scoring rules, which recover InfoNCE-anchor as a special case when the log score is employed. This formulation unifies a broad spectrum of contrastive objectives, including NCE, InfoNCE, and $f$-divergence variants, under a single principled framework. Empirically, we find that InfoNCE-anchor with the log score achieves the most accurate MI estimates; however, in self-supervised representation learning experiments, we find that the anchor does not improve the downstream task performance. These findings corroborate that contrastive representation learning benefits not from accurate MI estimation per se, but from the learning of structured density ratios.
format Preprint
id arxiv_https___arxiv_org_abs_2510_25983
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Contrastive Predictive Coding Done Right for Mutual Information Estimation
Ryu, J. Jon
Yeddanapudi, Pavan
Xu, Xiangxiang
Wornell, Gregory W.
Machine Learning
Information Theory
The InfoNCE objective, originally introduced for contrastive representation learning, has become a popular choice for mutual information (MI) estimation, despite its indirect connection to MI. In this paper, we demonstrate why InfoNCE should not be regarded as a valid MI estimator, and we introduce a simple modification, which we refer to as InfoNCE-anchor, for accurate MI estimation. Our modification introduces an auxiliary anchor class, enabling consistent density ratio estimation and yielding a plug-in MI estimator with significantly reduced bias. Beyond this, we generalize our framework using proper scoring rules, which recover InfoNCE-anchor as a special case when the log score is employed. This formulation unifies a broad spectrum of contrastive objectives, including NCE, InfoNCE, and $f$-divergence variants, under a single principled framework. Empirically, we find that InfoNCE-anchor with the log score achieves the most accurate MI estimates; however, in self-supervised representation learning experiments, we find that the anchor does not improve the downstream task performance. These findings corroborate that contrastive representation learning benefits not from accurate MI estimation per se, but from the learning of structured density ratios.
title Contrastive Predictive Coding Done Right for Mutual Information Estimation
topic Machine Learning
Information Theory
url https://arxiv.org/abs/2510.25983