Lab
SoftComputing and Data Analysis (SCoDA)
About the lab
SCoDA (SoftComputing and Data Analysis)
Data Mining, Pattern Recognition, Optimization, Agent technology, Fuzzy systems, Text mininig, GIS
Data Mining, Pattern Recognition, Optimization, Agent technology, Fuzzy systems, Text mininig, GIS
Featured research (1)
ABSTRACT:
Currently, password authentication is one of the most widely used security systems worldwide in many services. This results in the system security depends on the security of passwords, which, having to be easy to remember for humans, tend to be weak.
Gauges the strength of the currently defined passwords may return different estimates about the strength of the same password. This is due in part to the attacker used to crack the password, it has been shown that the strength of the pass-word depends on the type of attack and how to implement it by the attacker. And so, we cannot classify a mind-absolute in strong or weak pass-word, but must take into account a particular attack algorithm.
It aims to train a supervised classifier, to identify patterns that determine the features that make strong or weak password for a given attacker.
The results of this study allow to increase the security of password-protected systems. Sets policies acceptance or rejection of user pass-words and can generate an explanation to the user about why rejection of their password. Final-ly, these results allow administrators to constant-ly update their password security policies based on the latest versions of password attack sys-tems.
Keywords: security policies, password strength, identifying patterns
RESUMEN: En la actualidad, la autenticación por contraseña es uno de los sistemas de seguridad más utilizado mundialmente en muchos servicios. Esto trae como consecuencia que la seguridad de los siste-mas dependa de la seguridad de las contraseñas, las cuales, al tener que ser fáciles de recordar para los humanos, tienden a ser débiles. Los medidores de la fortaleza de las contraseñas definidos en la actualidad pueden devolver dife-rentes estimaciones sobre la fortaleza de la misma contraseña. Esto se debe, en parte al atacante que se utilice para romper la contraseña, pues se ha demostrado que la fortaleza de la contraseña dependerá del tipo de ataque y de la forma de im-plementarlo por parte del atacante. Por esto no se puede clasificar una contraseña absolutamente en fuerte o débil, sino que hay que tener en cuenta a un algoritmo de ataque en particular. Se propone entrenar un clasificador supervisado, que permita identificar patrones que determinen las características que hacen fuerte o débil una contraseña para un atacante dado. Los resultados de este trabajo permiten incremen-tar la seguridad de los sistemas protegidos por contraseña. Permite definir políticas de aceptación o rechazo de contraseñas de usuarios y puede ge-nerar una explicación al usuario sobre el porqué de rechazo de su contraseña. Finalmente, estos resultados permiten a los administradores la ac-tualización permanente de sus políticas de seguri-dad de contraseñas en función de las más recien-tes versiones de los sistemas de ataque de con-traseñas. Palabras claves: políticas de seguridad, forta-leza de contraseñas, identificación de patro-nes. ABSTRACT: Currently, password authentication is one of the most widely used security systems worldwide in many services. This results in the system security depends on the security of passwords, which, having to be easy to remember for humans, tend to be weak. Gauges the strength of the currently defined passwords may return different estimates about the strength of the same password. This is due in part to the attacker used to crack the password, it has been shown that the strength of the password depends on the type of attack and how to implement it by the attacker. And so, we cannot classify a mind-absolute in strong or weak password, but must take into account a particular attack algorithm. It aims to train a supervised classifier, to identify patterns that determine the features that make strong or weak password for a given attacker. The results of this study allow to increase the security of password-protected systems. Sets policies acceptance or rejection of user passwords and can generate an explanation to the user about why rejection of their password. Finally, these results allow administrators to constantly update their password security policies based on the latest versions of password attack systems.