Graph representations for reinforcement learning

Graph analysis is becoming increasingly important due to the expressive power of graph models and the efficient algorithms available for processing them. Reinforcement Learning is one domain that could benefit from advancements in graph analysis, given that a learning agent may be integrated into an...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Schab, Esteban, Casanova, Carlos, Piccoli, Fabiana
Formato: Articulo
Lenguaje:Inglés
Publicado: 2024
Materias:
Ciencias Informáticas
Computational Intelligence
Reinforcement Learning
Graph Embeddings
unsupervised GRL
Whole Graph Embedding
Inteligencia Computacional
Aprendizaje por Refuerzo
Embeddings de grafos
GRL no supervisado
Embedding de grafo entero
Acceso en línea:http://sedici.unlp.edu.ar/handle/10915/166757
Aporte de:
SEDICI (UNLP) de Universidad Nacional de La Plata
id I19-R120-10915-166757
record_format dspace
spelling I19-R120-10915-1667572024-06-05T20:04:19Z http://sedici.unlp.edu.ar/handle/10915/166757 Graph representations for reinforcement learning Representaciones de grafos para aprendizaje por refuerzo Schab, Esteban Casanova, Carlos Piccoli, Fabiana 2024-04 2024-06-05T13:28:56Z en Ciencias Informáticas Computational Intelligence Reinforcement Learning Graph Embeddings unsupervised GRL Whole Graph Embedding Inteligencia Computacional Aprendizaje por Refuerzo Embeddings de grafos GRL no supervisado Embedding de grafo entero Graph analysis is becoming increasingly important due to the expressive power of graph models and the efficient algorithms available for processing them. Reinforcement Learning is one domain that could benefit from advancements in graph analysis, given that a learning agent may be integrated into an environment that can be represented as a graph. Nevertheless, the structural irregularity of graphs and the lack of prior labels make it difficult to integrate such a model into modern Reinforcement Learning frameworks that rely on artificial neural networks. Graph embedding enables the learning of low-dimensional vector representations that are more suited for machine learning algorithms, while retaining essential graph features. This paper presents a framework for evaluating graph embedding algorithms and their ability to preserve the structure and relevant features of graphs by means of an internal validation metric, without resorting to subsequent tasks that require labels for training. Based on this framework, three defined algorithms that meet the necessary requirements for solving a specific problem of Reinforcement Learning in graphs are selected, analyzed, and compared. These algorithms are Graph2Vec, GL2Vec, and Wavelet Characteristics, with the latter two demonstrating superior performance. El análisis de grafos es un tópico emergente debido a la expresividad de los modelos basados en grafos y al desarrollo de algoritmos para su procesamiento. Un área que puede beneficiarse de estos avances es el aprendizaje por refuerzo, dado que un agente de aprendizaje puede estar imnerso en un entorno modelable como un grafo. Sin embargo, tanto la irregularidad de las características estructurales de los grafos como la ausencia de etiquetas a priori dificultan la incorporación de un modelo de este tipo en los marcos modernos de Aprendizaje por Refuerzo basados en redes neuronales artificiales. En este sentido, los embeddings de grafos permiten aprender representaciones vectoriales de baja dimensión, más adecuadas para los algoritmos de aprendizaje automático, preservando al mismo tiempo las características clave de los grafos. Proponemos un marco para evaluar algoritmos de Graph Embedding y su capacidad para preservar la estructura y características relevantes de los grafos mediante una métrica de validación interna, sin recurrir a tareas posteriores que requieran etiquetas para el entrenamiento. Aplicando este marco con un problema concreto, se seleccionan, analizan y comparan tres algoritmos que cumplen los requisitos necesarios: Graph2Vec, GL2Vec y Wavelet Characteristics, donde los dos últimos muestran un mejor comportamiento. Facultad de Informática Articulo Articulo http://creativecommons.org/licenses/by-nc-sa/4.0/ Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) application/pdf 29-38
institution Universidad Nacional de La Plata
institution_str I-19
repository_str R-120
collection SEDICI (UNLP)
language Inglés
topic Ciencias Informáticas
Computational Intelligence
Reinforcement Learning
Graph Embeddings
unsupervised GRL
Whole Graph Embedding
Inteligencia Computacional
Aprendizaje por Refuerzo
Embeddings de grafos
GRL no supervisado
Embedding de grafo entero
spellingShingle Ciencias Informáticas
Computational Intelligence
Reinforcement Learning
Graph Embeddings
unsupervised GRL
Whole Graph Embedding
Inteligencia Computacional
Aprendizaje por Refuerzo
Embeddings de grafos
GRL no supervisado
Embedding de grafo entero
Schab, Esteban
Casanova, Carlos
Piccoli, Fabiana
Graph representations for reinforcement learning
topic_facet Ciencias Informáticas
Computational Intelligence
Reinforcement Learning
Graph Embeddings
unsupervised GRL
Whole Graph Embedding
Inteligencia Computacional
Aprendizaje por Refuerzo
Embeddings de grafos
GRL no supervisado
Embedding de grafo entero
description Graph analysis is becoming increasingly important due to the expressive power of graph models and the efficient algorithms available for processing them. Reinforcement Learning is one domain that could benefit from advancements in graph analysis, given that a learning agent may be integrated into an environment that can be represented as a graph. Nevertheless, the structural irregularity of graphs and the lack of prior labels make it difficult to integrate such a model into modern Reinforcement Learning frameworks that rely on artificial neural networks. Graph embedding enables the learning of low-dimensional vector representations that are more suited for machine learning algorithms, while retaining essential graph features. This paper presents a framework for evaluating graph embedding algorithms and their ability to preserve the structure and relevant features of graphs by means of an internal validation metric, without resorting to subsequent tasks that require labels for training. Based on this framework, three defined algorithms that meet the necessary requirements for solving a specific problem of Reinforcement Learning in graphs are selected, analyzed, and compared. These algorithms are Graph2Vec, GL2Vec, and Wavelet Characteristics, with the latter two demonstrating superior performance.
format Articulo
Articulo
author Schab, Esteban
Casanova, Carlos
Piccoli, Fabiana
author_facet Schab, Esteban
Casanova, Carlos
Piccoli, Fabiana
author_sort Schab, Esteban
title Graph representations for reinforcement learning
title_short Graph representations for reinforcement learning
title_full Graph representations for reinforcement learning
title_fullStr Graph representations for reinforcement learning
title_full_unstemmed Graph representations for reinforcement learning
title_sort graph representations for reinforcement learning
publishDate 2024
url http://sedici.unlp.edu.ar/handle/10915/166757
work_keys_str_mv AT schabesteban graphrepresentationsforreinforcementlearning
AT casanovacarlos graphrepresentationsforreinforcementlearning
AT piccolifabiana graphrepresentationsforreinforcementlearning
AT schabesteban representacionesdegrafosparaaprendizajeporrefuerzo
AT casanovacarlos representacionesdegrafosparaaprendizajeporrefuerzo
AT piccolifabiana representacionesdegrafosparaaprendizajeporrefuerzo
_version_ 1807223369885548544

Ejemplares similares