\(\DeclarePairedDelimiterX{\Set}[2]{\{}{\}}{#1 \nonscript\;\delimsize\vert\nonscript\; #2}\) \( \DeclarePairedDelimiter{\set}{\{}{\}}\) \( \DeclarePairedDelimiter{\parens}{\left(}{\right)}\) \(\DeclarePairedDelimiterX{\innerproduct}[1]{\langle}{\rangle}{#1}\) \(\newcommand{\ip}[1]{\innerproduct{#1}}\) \(\newcommand{\bmat}[1]{\left[\hspace{2.0pt}\begin{matrix}#1\end{matrix}\hspace{2.0pt}\right]}\) \(\newcommand{\barray}[1]{\left[\hspace{2.0pt}\begin{matrix}#1\end{matrix}\hspace{2.0pt}\right]}\) \(\newcommand{\mat}[1]{\begin{matrix}#1\end{matrix}}\) \(\newcommand{\pmat}[1]{\begin{pmatrix}#1\end{pmatrix}}\) \(\newcommand{\mathword}[1]{\mathop{\textup{#1}}}\)
Needs:
Parameterized Distributions
Latent Generation Pairs
Neural Networks
Needed by:
Variational Autoencoders
Links:
Sheet PDF
Graph PDF

Neural Distribution Families

Why

If neural networks are flexible function approximators, and we want flexible parameterizations of conditional distributions, we might use a neural network for the parameterizers.1

Definition

A neural conditional distribution family (or deep conditional distribution family) is a parameterized conditional distribution family whose parameterizer is a neural network (see Parameterized Distributions and Neural Networks). When clear from context, we shorten to neural conditional family (or deep conditional family).

We similarly define a neural latent generation family (or deep latent generation family) and more generally for neural distribution graphs (or deep distribution graphs). Other terminology for deep latent generation family includes deep latent variable model (DLVM).


  1. Future editions will expand and modify. ↩︎
Copyright © 2023 The Bourbaki Authors — All rights reserved — Version 13a6779cc About Show the old page view