---
title: "BAND: Bayesian Network Distribution Regression for High-Dimensional Distribution Estimation – Breaking the Curse"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.26955v1"
permalink: "https://feed.astralor.com/items/itm_ee1fa10527efbe75d764c46c"
published: "2026-07-29T14:21:39.000Z"
score: 85
channel: "ai"
tags: ["BAND","Sparsity","Bayesian networks","Nonparametric Density Estimation","High-Dimensional Distribution Estimation"]
---

# BAND: Bayesian Network Distribution Regression for High-Dimensional Distribution Estimation – Breaking the Curse

## AI commentary

> This paper is a must-read for anyone working on high-dimensional statistics, probabilistic modeling, and generative AI. It provides a simple yet powerful theoretical answer: sparsity can break the curse of dimensionality, and Bayesian networks offer a natural framework. Compared to black-box methods like VAEs and GANs, BAND is interpretable and provably convergent, especially valuable for industry needs like reliable confidence intervals. If you research time series forecasting or risk management, this paper may change how you approach high-dimensional density estimation.

## At a glance

Traditional high-dimensional distribution estimation suffers from the 'curse of dimensionality', with optimal convergence rates decaying exponentially. This paper proposes BAND (BAyesian Network Distribution regression), which leverages sparse Bayesian network structure and sparsity-aware conditional mean estimation to achieve polynomial total variation convergence rates while allowing feature dimension to grow polynomially with sample size. Empirical evaluations show BAND outperforms classical histogram density estimators in data sampling and confidence region forecasting tasks, offering a new paradigm for high-dimensional mixed-type time series modeling.

## Key points

- Proposes BAND, combining sparse Bayesian networks and conditional mean estimation, achieving polynomial total variation convergence in high-dimensional distribution estimation.
- Theoretically proves that under sparsity, feature dimension can grow polynomially with sample size, dramatically faster than classical exponential decay rates.
- Supports mixed data types and high-dimensional time series, outperforming multiple state-of-the-art baselines in data sampling and confidence region forecasting.
- Provides a scalable sparse modeling framework for high-dimensional density estimation, with theoretical insights guiding future nonparametric methods.

## Full text

Minimax-optimal rates for multivariate distribution estimation are known to suffer from the curse of dimensionality. We propose a sparse Bayesian network approach in which each conditional probability is estimated using sparsity-aware conditional mean methods. The resulting estimator, \textit{BAyesian Network Distribution regression} (BAND), handles mixed data types in high-dimensional time series and achieves polynomial total variation convergence rates while allowing the feature dimension to grow polynomially with the sample size. These rates are substantially faster than the classical optimal rates for multivariate histogram density estimators that lack sparsity. Empirical evaluations show that BAND performs competitively for data sampling and confidence region forecasting against a range of state-of-the-art benchmarks.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_ee1fa10527efbe75d764c46c