GFlowNet Foundations

TL;DR

GFlowNets are a versatile generative framework with theoretical properties enabling distribution estimation, efficient sampling, and extensions to complex, stochastic, and continuous environments, surpassing traditional methods like MCMC.

Contribution

This paper advances GFlowNets by establishing their ability to estimate complex distributions, introduce new variations, and extend their applicability to diverse, challenging settings.

Findings

GFlowNets can estimate joint and marginal distributions over complex objects.

They can efficiently approximate partition functions and free energies.

Extensions include entropy estimation, Pareto sampling, and handling stochastic environments.

Abstract

Generative Flow Networks (GFlowNets) have been introduced as a method to sample a diverse set of candidates in an active learning context, with a training objective that makes them approximately sample in proportion to a given reward function. In this paper, we show a number of additional theoretical properties of GFlowNets. They can be used to estimate joint probability distributions and the corresponding marginal distributions where some variables are unspecified and, of particular interest, can represent distributions over composite objects like sets and graphs. GFlowNets amortize the work typically done by computationally expensive MCMC methods in a single but trained generative pass. They could also be used to estimate partition functions and free energies, conditional probabilities of supersets (supergraphs) given a subset (subgraph), as well as marginal distributions over all supersets (supergraphs) of a given set (graph). We introduce variations enabling the estimation of entropy and mutual information, sampling from a Pareto frontier, connections to reward-maximizing policies, and extensions to stochastic environments, continuous actions and modular energy functions.