Predicting recurring concepts on data-streams by me ans of a meta-model and a fuzzy similarity function

Abad Arranz, Miguel Ángel; Gomes, João Bartolo y Menasalvas Ruiz, Ernestina (2015). Predicting recurring concepts on data-streams by me ans of a meta-model and a fuzzy similarity function. "Expert Systems With Applications", v. 46 ; pp. 87-105. ISSN 0957-4174. https://doi.org/10.1016/j.eswa.2015.10.022.

Descripción

Título: Predicting recurring concepts on data-streams by me ans of a meta-model and a fuzzy similarity function
Autor/es:
  • Abad Arranz, Miguel Ángel
  • Gomes, João Bartolo
  • Menasalvas Ruiz, Ernestina
Tipo de Documento: Artículo
Título de Revista/Publicación: Expert Systems With Applications
Fecha: Marzo 2015
Volumen: 46
Materias:
Palabras Clave Informales: Data stream mining; Hidden Markov model; Fuzzy logic; Concept drift; Classification
Escuela: E.T.S. de Ingenieros Informáticos (UPM)
Departamento: Lenguajes y Sistemas Informáticos e Ingeniería del Software
Licencias Creative Commons: Reconocimiento - Sin obra derivada - No comercial

Texto completo

[img]
Vista Previa
PDF (Document Portable Format) - Se necesita un visor de ficheros PDF, como GSview, Xpdf o Adobe Acrobat Reader
Descargar (1MB) | Vista Previa

Resumen

Meta-models can be used in the process of enhancing the drift detection mechanisms used by data stream algorithms, by representing and predicting when the change will occur. There are some real-world situations where a concept reappears, as in the case of intrusion detection systems(IDS), where the same incidents or an adaptation of them usually reappear over time. In these environments the early prediction of drift by means of a better knowledge of past models can help to anticipate to the change, thus improving efficiency of the model regarding the training instances needed. In this paper we present MM-PRec, a meta-model for predicting recurring concepts on data-streams which main goal is to predict when the drift is going to occur together with the best model to be used in case of a recurring concept. To fulfill this goal, MM-PRec trains a Hidden Markov Model (HMM) from the instances that appear during the concept drift. The learning process of the base classification learner feeds the meta-model with all the information needed to predict recurrent or similar situations. Thus, the models predicted together with the associated contextual information are stored. In our approach we also propose to use a fuzzy similarity function to decide which is the best model to represent a particular context when drift is detected. The experiments performed show that MM-PRec outperforms the behaviour of other context-aware algorithms in terms of training instances needed, specially in environments characterized by the presence of gradual drifts.

Más información

ID de Registro: 43634
Identificador DC: http://oa.upm.es/43634/
Identificador OAI: oai:oa.upm.es:43634
Identificador DOI [BETA]: 10.1016/j.eswa.2015.10.022
URL Oficial: http://www.sciencedirect.com/science/article/pii/S0957417415007174
Depositado por: Memoria Investigacion
Depositado el: 24 Mar 2017 12:45
Ultima Modificación: 24 Mar 2017 13:23
  • Open Access
  • Open Access
  • Sherpa-Romeo
    Compruebe si la revista anglosajona en la que ha publicado un artículo permite también su publicación en abierto.
  • Dulcinea
    Compruebe si la revista española en la que ha publicado un artículo permite también su publicación en abierto.
  • Recolecta
  • e-ciencia
  • Observatorio I+D+i UPM
  • OpenCourseWare UPM