@InProceedings{BarchiHrusCostCarv:2016:GaOnEx,
author = "Barchi, Paulo and Hruschka Junior, Estevam and Costa, Fausto Guzzo
da and Carvalho, Reinaldo Ramos de",
affiliation = "{} and {} and {} and {Instituto Nacional de Pesquisas Espaciais
(INPE)}",
title = "Galaxies ontology extension through deep learning",
year = "2016",
organization = "Workshop de Computa{\c{c}}{\~a}o Aplicada, 16. (WORCAP)",
abstract = "Ontology Extension is a Machine Learning (ML) technique commonly
used to expand Knowledge Bases (KB). One approach in Machine
Reading (MR) is to identify and add to the KB new relations that
are frequently asserted in huge text data. These huge amount of
data (not necessarily text) can also be referred as data hypercube
because of its simple data structure and logic which represent the
data in the (multiple) dimensions of interest. Co-occurrence
matrices are used to structure the normalized values of
co-occurrence between the contexts for each category pair to
identify those context patterns. After the clustering phase, from
each cluster arises a new possible relation. This work presents a
new application to use this approach to expand the Ontology of
Galaxies. Convolution Neural Networks (CNN) are deep neural
networks appropriate to handle images. The main idea is to train,
test and validate one CNN connected to a Support Vector Machine
(SVM) - well known for its strong theoretical base and practical
effiency - from Galaxy Zoo data warehousing; and apply this system
(CNN with SVM) to classify new galaxies never seen before from the
SLOAN database, and thus, to extend the ontology of galaxies.",
conference-location = "S{\~a}o Jos{\'e} dos Campos, SP",
conference-year = "25-26 out.",
language = "en",
targetfile = "barchi_galaxies.pdf",
urlaccessdate = "27 abr. 2024"
}