The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
O objetivo da incorporação de gráficos é aprender uma função de incorporação de dimensão inferior para dados gráficos. Os métodos existentes geralmente dependem da estimativa de máxima verossimilhança (MLE) e muitas vezes aprendem uma função de incorporação por meio da estimativa de média condicional (CME). No entanto, o MLE é conhecido por ser vulnerável à contaminação de valores discrepantes. Além disso, o CME pode restringir a aplicabilidade dos métodos de incorporação de gráficos a uma gama limitada de dados gráficos. Para lidar com esses problemas, este artigo propõe um novo método para incorporação de grafos denominado incorporação de gráfico de proporção robusta (RRG). RRGE é baseado na estimativa da razão entre as distribuições de probabilidade condicional e marginal dos pesos dos links dados vetores de dados e seria aplicável a uma gama mais ampla de dados gráficos do que os métodos baseados em CME. Além disso, para obter uma estimativa robusta de outliers, a razão é estimada com a entropia cruzada γ, que é uma alternativa robusta à entropia cruzada padrão. Experimentos numéricos em dados artificiais mostram que o RRGE é robusto contra valores discrepantes e tem um bom desempenho mesmo quando os métodos baseados em CME não funcionam de todo. Finalmente, o desempenho do método proposto é demonstrado em conjuntos de dados do mundo real utilizando redes neurais.
Kaito SATTA
Future University Hakodate
Hiroaki SASAKI
Future University Hakodate
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Kaito SATTA, Hiroaki SASAKI, "Graph Embedding with Outlier-Robust Ratio Estimation" in IEICE TRANSACTIONS on Information,
vol. E105-D, no. 10, pp. 1812-1816, October 2022, doi: 10.1587/transinf.2022EDL8033.
Abstract: The purpose of graph embedding is to learn a lower-dimensional embedding function for graph data. Existing methods usually rely on maximum likelihood estimation (MLE), and often learn an embedding function through conditional mean estimation (CME). However, MLE is well-known to be vulnerable to the contamination of outliers. Furthermore, CME might restrict the applicability of the graph embedding methods to a limited range of graph data. To cope with these problems, this paper proposes a novel method for graph embedding called the robust ratio graph embedding (RRGE). RRGE is based on the ratio estimation between the conditional and marginal probability distributions of link weights given data vectors, and would be applicable to a wider-range of graph data than CME-based methods. Moreover, to achieve outlier-robust estimation, the ratio is estimated with the γ-cross entropy, which is a robust alternative to the standard cross entropy. Numerical experiments on artificial data show that RRGE is robust against outliers and performs well even when CME-based methods do not work at all. Finally, the performance of the proposed method is demonstrated on realworld datasets using neural networks.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2022EDL8033/_p
Copiar
@ARTICLE{e105-d_10_1812,
author={Kaito SATTA, Hiroaki SASAKI, },
journal={IEICE TRANSACTIONS on Information},
title={Graph Embedding with Outlier-Robust Ratio Estimation},
year={2022},
volume={E105-D},
number={10},
pages={1812-1816},
abstract={The purpose of graph embedding is to learn a lower-dimensional embedding function for graph data. Existing methods usually rely on maximum likelihood estimation (MLE), and often learn an embedding function through conditional mean estimation (CME). However, MLE is well-known to be vulnerable to the contamination of outliers. Furthermore, CME might restrict the applicability of the graph embedding methods to a limited range of graph data. To cope with these problems, this paper proposes a novel method for graph embedding called the robust ratio graph embedding (RRGE). RRGE is based on the ratio estimation between the conditional and marginal probability distributions of link weights given data vectors, and would be applicable to a wider-range of graph data than CME-based methods. Moreover, to achieve outlier-robust estimation, the ratio is estimated with the γ-cross entropy, which is a robust alternative to the standard cross entropy. Numerical experiments on artificial data show that RRGE is robust against outliers and performs well even when CME-based methods do not work at all. Finally, the performance of the proposed method is demonstrated on realworld datasets using neural networks.},
keywords={},
doi={10.1587/transinf.2022EDL8033},
ISSN={1745-1361},
month={October},}
Copiar
TY - JOUR
TI - Graph Embedding with Outlier-Robust Ratio Estimation
T2 - IEICE TRANSACTIONS on Information
SP - 1812
EP - 1816
AU - Kaito SATTA
AU - Hiroaki SASAKI
PY - 2022
DO - 10.1587/transinf.2022EDL8033
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E105-D
IS - 10
JA - IEICE TRANSACTIONS on Information
Y1 - October 2022
AB - The purpose of graph embedding is to learn a lower-dimensional embedding function for graph data. Existing methods usually rely on maximum likelihood estimation (MLE), and often learn an embedding function through conditional mean estimation (CME). However, MLE is well-known to be vulnerable to the contamination of outliers. Furthermore, CME might restrict the applicability of the graph embedding methods to a limited range of graph data. To cope with these problems, this paper proposes a novel method for graph embedding called the robust ratio graph embedding (RRGE). RRGE is based on the ratio estimation between the conditional and marginal probability distributions of link weights given data vectors, and would be applicable to a wider-range of graph data than CME-based methods. Moreover, to achieve outlier-robust estimation, the ratio is estimated with the γ-cross entropy, which is a robust alternative to the standard cross entropy. Numerical experiments on artificial data show that RRGE is robust against outliers and performs well even when CME-based methods do not work at all. Finally, the performance of the proposed method is demonstrated on realworld datasets using neural networks.
ER -