Generative Flow Networks for Discrete Probabilistic Modeling
This addresses probabilistic modeling challenges for researchers working with discrete data, representing an incremental advancement that combines GFlowNets with energy-based methods.
The paper tackles probabilistic modeling for high-dimensional discrete data by introducing energy-based generative flow networks (EB-GFN), which jointly train a GFlowNet to sample from an energy distribution and an energy function with an approximate maximum likelihood objective. The method demonstrates effectiveness on various tasks, though no concrete performance numbers are provided in the abstract.
We present energy-based generative flow networks (EB-GFN), a novel probabilistic modeling algorithm for high-dimensional discrete data. Building upon the theory of generative flow networks (GFlowNets), we model the generation process by a stochastic data construction policy and thus amortize expensive MCMC exploration into a fixed number of actions sampled from a GFlowNet. We show how GFlowNets can approximately perform large-block Gibbs sampling to mix between modes. We propose a framework to jointly train a GFlowNet with an energy function, so that the GFlowNet learns to sample from the energy distribution, while the energy learns with an approximate MLE objective with negative samples from the GFlowNet. We demonstrate EB-GFN's effectiveness on various probabilistic modeling tasks. Code is publicly available at https://github.com/zdhNarsil/EB_GFN.