Pessimistic Backward Policy for GFlowNets

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Jang, Hyosoon, Jang, Yunhui, Kim, Minsu, Park, Jinkyoo, Ahn, Sungsoo
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866913566671503360
author Jang, Hyosoon
Jang, Yunhui
Kim, Minsu
Park, Jinkyoo
Ahn, Sungsoo
author_facet Jang, Hyosoon
Jang, Yunhui
Kim, Minsu
Park, Jinkyoo
Ahn, Sungsoo
contents This paper studies Generative Flow Networks (GFlowNets), which learn to sample objects proportionally to a given reward function through the trajectory of state transitions. In this work, we observe that GFlowNets tend to under-exploit the high-reward objects due to training on insufficient number of trajectories, which may lead to a large gap between the estimated flow and the (known) reward value. In response to this challenge, we propose a pessimistic backward policy for GFlowNets (PBP-GFN), which maximizes the observed flow to align closely with the true reward for the object. We extensively evaluate PBP-GFN across eight benchmarks, including hyper-grid environment, bag generation, structured set generation, molecular generation, and four RNA sequence generation tasks. In particular, PBP-GFN enhances the discovery of high-reward objects, maintains the diversity of the objects, and consistently outperforms existing methods.
format Preprint
id arxiv_https___arxiv_org_abs_2405_16012
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Pessimistic Backward Policy for GFlowNets
Jang, Hyosoon
Jang, Yunhui
Kim, Minsu
Park, Jinkyoo
Ahn, Sungsoo
Machine Learning
This paper studies Generative Flow Networks (GFlowNets), which learn to sample objects proportionally to a given reward function through the trajectory of state transitions. In this work, we observe that GFlowNets tend to under-exploit the high-reward objects due to training on insufficient number of trajectories, which may lead to a large gap between the estimated flow and the (known) reward value. In response to this challenge, we propose a pessimistic backward policy for GFlowNets (PBP-GFN), which maximizes the observed flow to align closely with the true reward for the object. We extensively evaluate PBP-GFN across eight benchmarks, including hyper-grid environment, bag generation, structured set generation, molecular generation, and four RNA sequence generation tasks. In particular, PBP-GFN enhances the discovery of high-reward objects, maintains the diversity of the objects, and consistently outperforms existing methods.
title Pessimistic Backward Policy for GFlowNets
topic Machine Learning
url https://arxiv.org/abs/2405.16012