The Master dissertation of Leonardo Filipe Rodrigues Ribeiro entitled "Learning Network Node Representations from Structural Identity" and supervised by Daniel Ratton Figueiredo, received first place in the  11th Competition of Theses and Dissertations in Artificial and Computational Intelligence (CTDIAC), a satellite event of the Brazilian Conference on Intelligent Systems (BRACIS), held in São Paulo from October 22 to 25, 2018.


The dissertation presents a new methodology for embedding network nodes in Euclidean space in order to preserve their structural similarity, and finds applications in the discovery of roles for network nodes and node classification problems. The work has been published in ACM SIGKDD 2017 (view news) and was well received by the community, having already accumulated 70 according to Google Scholar.


The 11th CTDIAC received 25 master dissertations (and 19 doctoral theses) in its first phase, from which four were selected to the second phase which consists of an oral presentation and questioning by the organizers (view list of final four). In addition to Leonardo, the master thesis entitled "Speeding Up Parameter and Rule Learning for Acyclic Probabilistic Logic Programs "by Francisco Faria, supervised by Fabio Cozman (USP), also received the first place.


Congratulations to Leonardo and his advisor for the achievement!




(1) Award Certificate:


(2) Award ceremony (from left to right): Richard Aderbal Gonçalves (CTDIAC organizer), Leonardo Filipe Rodrigues Ribeiro (winner from PESC), Francisco Faria (winner from USP), Ronaldo Cristiano Prati (CTDIAC organizer):