rogergranada,Roger Granada,github

Hi there, I'm Roger 👨‍💻

I am graduated in Systems Development and Analysis from the Catholic University of Pelotas (UCPel), Brazil. During my under graduation, I worked with research in recommender systems at the System Information Research Group (GPSI) using text mining to build user's profile to recommender systems. In 2009, I joined the Hewlett Packard/PUCRS Research Centre in Automating Privacy Assessments using Ontologies, in cooperation with the HP Labs from Princeton. In that project I worked with automatic thesaurus construction and ontologies on the privacy domain. Two years later I received the Master's degree in Computer Science from Pontifical Catholic University of Rio Grande do Sul (PUCRS) for a thesis involving automatic thesaurus construction.

Click here to read more...

In 2011 at PUCRS, I started the PhD focusing the research on the discovery of the word meaning, using the thesaurus structure to get it. During the second year of the PhD I did an internship at Fondazione Bruno Kessler ([FBK-irst](hlt.fbk.eu)) in Trento (Italy), where I worked with Cross-Lingual Distributional Similarity Models. A year later, I was a research visitor and PhD student at Institut de Recherche en Informatique de Toulouse ([IRIT](www.irit.fr)) in Toulouse (France). In a cotutelle agreement between PUCRS and Université Toulouse III - Paul Sabatier ([UPS3](http://www.univ-tlse3.fr/), I finished the PhD in 2015 with a thesis about the automatic construction of hierarchical relations from plain texts.

In 2016, as a postdoctoral researcher, I joined a Hewlett Packard/PUCRS project about the identification of action and goals in video sequences. In that project I started working with deep learning algorithms and computer vision. A year later, I receive a postdoctoral scholarship from CAPES to work in the Pró-Alertas project which aims at builging an integrated system to control a team of heterogeneous robots (aerial, boats, land) and alert systems. In 2018, I received postdoctoral scholarship from CAPES/FAPERGS DocFix to work with goal and plan recognition using real world data. The project uses videos of human beings performing actions in a kitchen to discover the recipe their are doing. It uses deep learning algorithms to recognize objects and the relationships between them, and then, infer the plan the subject is pursuing based on the objects and relationships recognized.

For a complete description of my activities, check out my Page or my curriculum Lattes. For a brief description, check out my CV