Distributional GFlowNets with Quantile Flows

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Zhang, Dinghuai, Pan, Ling, Chen, Ricky T. Q., Courville, Aaron, Bengio, Yoshua
Formato: Preprint
Publicado: 2023
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909109890056192
author Zhang, Dinghuai
Pan, Ling
Chen, Ricky T. Q.
Courville, Aaron
Bengio, Yoshua
author_facet Zhang, Dinghuai
Pan, Ling
Chen, Ricky T. Q.
Courville, Aaron
Bengio, Yoshua
contents Generative Flow Networks (GFlowNets) are a new family of probabilistic samplers where an agent learns a stochastic policy for generating complex combinatorial structure through a series of decision-making steps. Despite being inspired from reinforcement learning, the current GFlowNet framework is relatively limited in its applicability and cannot handle stochasticity in the reward function. In this work, we adopt a distributional paradigm for GFlowNets, turning each flow function into a distribution, thus providing more informative learning signals during training. By parameterizing each edge flow through their quantile functions, our proposed \textit{quantile matching} GFlowNet learning algorithm is able to learn a risk-sensitive policy, an essential component for handling scenarios with risk uncertainty. Moreover, we find that the distributional approach can achieve substantial improvement on existing benchmarks compared to prior methods due to our enhanced training algorithm, even in settings with deterministic rewards.
format Preprint
id arxiv_https___arxiv_org_abs_2302_05793
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Distributional GFlowNets with Quantile Flows
Zhang, Dinghuai
Pan, Ling
Chen, Ricky T. Q.
Courville, Aaron
Bengio, Yoshua
Machine Learning
Artificial Intelligence
Computation
Generative Flow Networks (GFlowNets) are a new family of probabilistic samplers where an agent learns a stochastic policy for generating complex combinatorial structure through a series of decision-making steps. Despite being inspired from reinforcement learning, the current GFlowNet framework is relatively limited in its applicability and cannot handle stochasticity in the reward function. In this work, we adopt a distributional paradigm for GFlowNets, turning each flow function into a distribution, thus providing more informative learning signals during training. By parameterizing each edge flow through their quantile functions, our proposed \textit{quantile matching} GFlowNet learning algorithm is able to learn a risk-sensitive policy, an essential component for handling scenarios with risk uncertainty. Moreover, we find that the distributional approach can achieve substantial improvement on existing benchmarks compared to prior methods due to our enhanced training algorithm, even in settings with deterministic rewards.
title Distributional GFlowNets with Quantile Flows
topic Machine Learning
Artificial Intelligence
Computation
url https://arxiv.org/abs/2302.05793