Optimizing Backward Policies in GFlowNets via Trajectory Likelihood Maximization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Gritsaev, Timofei, Morozov, Nikita, Samsonov, Sergey, Tiapkin, Daniil
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909522049630208
author Gritsaev, Timofei
Morozov, Nikita
Samsonov, Sergey
Tiapkin, Daniil
author_facet Gritsaev, Timofei
Morozov, Nikita
Samsonov, Sergey
Tiapkin, Daniil
contents Generative Flow Networks (GFlowNets) are a family of generative models that learn to sample objects with probabilities proportional to a given reward function. The key concept behind GFlowNets is the use of two stochastic policies: a forward policy, which incrementally constructs compositional objects, and a backward policy, which sequentially deconstructs them. Recent results show a close relationship between GFlowNet training and entropy-regularized reinforcement learning (RL) problems with a particular reward design. However, this connection applies only in the setting of a fixed backward policy, which might be a significant limitation. As a remedy to this problem, we introduce a simple backward policy optimization algorithm that involves direct maximization of the value function in an entropy-regularized Markov Decision Process (MDP) over intermediate rewards. We provide an extensive experimental evaluation of the proposed approach across various benchmarks in combination with both RL and GFlowNet algorithms and demonstrate its faster convergence and mode discovery in complex environments.
format Preprint
id arxiv_https___arxiv_org_abs_2410_15474
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Optimizing Backward Policies in GFlowNets via Trajectory Likelihood Maximization
Gritsaev, Timofei
Morozov, Nikita
Samsonov, Sergey
Tiapkin, Daniil
Machine Learning
Generative Flow Networks (GFlowNets) are a family of generative models that learn to sample objects with probabilities proportional to a given reward function. The key concept behind GFlowNets is the use of two stochastic policies: a forward policy, which incrementally constructs compositional objects, and a backward policy, which sequentially deconstructs them. Recent results show a close relationship between GFlowNet training and entropy-regularized reinforcement learning (RL) problems with a particular reward design. However, this connection applies only in the setting of a fixed backward policy, which might be a significant limitation. As a remedy to this problem, we introduce a simple backward policy optimization algorithm that involves direct maximization of the value function in an entropy-regularized Markov Decision Process (MDP) over intermediate rewards. We provide an extensive experimental evaluation of the proposed approach across various benchmarks in combination with both RL and GFlowNet algorithms and demonstrate its faster convergence and mode discovery in complex environments.
title Optimizing Backward Policies in GFlowNets via Trajectory Likelihood Maximization
topic Machine Learning
url https://arxiv.org/abs/2410.15474