Probabilistic Neural Circuits

Pedro Zuidberg Dos Martires
3/10/2024
cs.LGcs.AIcs.NE

No official code found. Best implementation: pedrozudo/ProbabilisticNeuralCircuits (6 stars, Apache-2.0, last commit 2024-04-22).

Abstract

Probabilistic circuits (PCs) have gained prominence in recent years as a versatile framework for discussing probabilistic models that support tractable queries and are yet expressive enough to model complex probability distributions. Nevertheless, tractability comes at a cost: PCs are less expressive than neural networks. In this paper we introduce probabilistic neural circuits (PNCs), which strike a balance between PCs and neural nets in terms of tractability and expressive power. Theoretically, we show that PNCs can be interpreted as deep mixtures of Bayesian networks. Experimentally, we demonstrate that PNCs constitute powerful function approximators.

View on arXivView PDF

Code Implementations(1)

pedrozudo/ProbabilisticNeuralCircuitsCommunity100%
Apache-2.0
60PythonDec 14, 20232 years agoApache-2.0

Similar Papers

Building Expressive and Tractable Probabilistic Generative Models: A Review

87% match2024

Tractable Representation Learning with Probabilistic Circuits

86% match1 repo12025

Neural Probabilistic Circuits: Enabling Compositional and Interpretable Predictions through Logical Reasoning

85% match2025

Probabilistic Integral Circuits

85% match2023

On Neural Networks as Infinite Tree-Structured Probabilistic Graphical Models

83% match2023

On Neural Networks as Infinite Tree-Structured Probabilistic Graphical Models

83% match2026

Cite this paper

@article{martires2024probabilistic,
  title  = {Probabilistic Neural Circuits},
  author = {Pedro Zuidberg Dos Martires},
  year   = {2024},
  eprint = {2403.06235},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2403.06235v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2403.06235.svg)](https://codeofpaper.com/paper/2403.06235)

Discussion