Advanced search
Start date

Dynamic semi-supervised and active learning based on complex networks


The main purpose of this project is the development of new techniques for semi-supervised learning-based networks for dynamic data sets. Properties of complex networks that represent the data and dynamic computational models for label propagation will be taken into account. Measures of complex networks will be extracted and used as parameters for selection of vertices. This selection is two-fold: indicate which are the best samples for labeling (active learning); and which vertices of the network attach the new instances (dynamic network). The label propagation in the network (semi-supervised classification) is performed by dynamic computational models, focusing on particle competition and neuronal synchronization models. From the study conducted during this project and the development of new techniques, we expected to generate original contributions in three main fields: 1) representation of dynamic datasets in networks; 2) development of techniques capable of dealing with dynamic data; and 3) active learning based on properties of complex networks to optimize the annotator work. This project consists on the continuation and expansion of the post-doctoral project, Proc. Fapesp no. 2008/09553-4, interrupted in November of 2009 owning to the position of the researcher as a professor at ICT/UNIFESP. (AU)

Articles published in Agência FAPESP Newsletter about the research grant:
Articles published in other media outlets (0 total):
More itemsLess items

Scientific publications
(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)
QUILES, MARCOS G.; MACAU, ELBERT E. N.; RUBIDO, NICOLAS. Dynamical detection of network communities. SCIENTIFIC REPORTS, v. 6, . (11/18496-7, 11/50151-0, 15/50122-0)
BREVE, FABRICIO A.; ZHAO, LIANG; QUILES, MARCOS G.. Particle competition and cooperation for semi-supervised learning with label noise. Neurocomputing, v. 160, p. 63-72, . (11/18496-7, 11/50151-0, 11/17396-9, 13/07375-0)

Please report errors in scientific publications list by writing to: