Ricardo Matsumura Araujo
CS Professor at UFPel

About me

CS Professor at the Center for Technological Advancement (CDTec) in the Federal University of Pelotas (UFPel). PhD and Master Degrees in Computer Science (from UFRGS). Lecturing at the Computer Science, Computer Engineering undergrad courses and the Graduation Program in Computer Science.

Machine Learning. Computational Social Science. Data Analysis. Web Development.

Twitter Linkedin CV Lattes E-mail
Ricardo's Picture

Blog random ramblings (often in portuguese)

Continue reading RSS feed

Projects research and otherwise

Biometric Identification using Kinect sensors

We aim at using the popular Microsoft Kinect device to passively capture biometric data other than face recognition and use it for robust person identification. We are currently looking into using anthropometric information, both from static and dynamic scenarios (walks, gestures).

Students involved in this project:
  • Diego Jaccottet (MSc student)
  • Leandro Dias(MSc student)
  • Rafael Dutra (Comp.Eng. student)
Former students involved:
  • Virginia Andersson (MSc student)
  • Gustavo Graña (CS student)
More about this
Kinect device

Social Information Processing

Product of my PhD thesis, Memetic Networks are a set of search heuristic based on a model of how people consume and communicate information. It uses the concept of memes and how they are exchanged, combined and modified through communication and reasoning over a social network. While initially proposed as a model to experiment with network properties and their effect on problem-solving performance, Memetic Networks were shown to be competitive with other search heuristics in some optimization tasks. This project is partly funded by FAPERGS and CNPq.

Other researchers involved:
  • Luis Lamb
  • Diego Noble (MS student at UFRGS)
  • Daniel Farenzena (PhD student at UFRGS)
Papers on this:
More about this
A Graph

Microcontent Priorization and Recommendation

We aim at creating a priority inbox for Twitter by learning the users' preferences and automatically ranking incoming tweets, surfacing those considered important. Moreover, we consider ways to inject relevant tweets from not-followed users (or even external sources) into one's stream. The challenge is leveraging social attributes and other meta-information while making the best use of a very limited amount of textual information. This project was funded by Bolsa Pesquisa UOL and is largely inactive since then (Twitter has been rolling some of these features to the end users).

Former students involved:
  • Tiago Schenkel (MSc student)
  • Vinicius Pazzini (CS student)
  • Mikael Poetsch (CS student)
  • Gabriel Siedler (CS student)
  • Gabriel Cardoso (CS student)
More about this
Twitter Composition

Publications recent and/or noteworthy

Person Identification Using Anthropometric and Gait Data from Kinect Sensor. Virginia Andersson, Ricardo Araujo. 29th AAAI Conference. 2015.
Abstract. Paper. Data set.
99designs: An Analysis of Creative Competition in Crowdsourced Design. Ricardo Araujo. AAAI Conference on Human Computation and Crowdsourcing (HComp). 2013.
Abstract. Paper. Data set.
Towards skeleton biometric identification using the Microsoft Kinect sensor. Ricardo Araujo, Virginia Andersson, Gustavo Graña. ACM Symposium on Applied Computing (SAC) 2013.
Abstract. Paper.
Full list (Lattes) Papers in Google Scholar
How Does Social Capital Affect Retweets?. Raquel Recuero, Ricardo Araujo, Gabriela Zago. AAAI ICWSM 2011.
Abstract. Paper.
Diffusion dynamics of games on online social networks. Xiao Wei, Jiang Yang, Lada Adamic, Ricardo Araujo, Manu Rekhi. USENIX WOSN 2010.
Abstract. Paper. Draft.
On the use of memory and resources in minority games. Ricardo Araujo, Luis Lamb. ACM Transactions on Autonomous and Adaptive Systems, 2009.
Abstract. Paper.

Courses recent only

Machine Learning

MS level at PPGC-UFPel (typically on second semesters). This course provides an introduction to Machine Learning, common algorithms, validation methodologies and a bit of computational learning theory. It is very project-oriented, requiring students to implement a few and test several algorithms on different scenarios. Students must know beforehand some popular programming language (Python is preferred) or an appropriate environment (R, Matlab). Knowledge of probability and statistics is a plus.

Algorithms and Data Structures

Offered to Computer Science and Computer Engineer students at UFPel. It covers data structures (B-Trees, Tries, Hash Tables, Graphs), algorithm design (dynamic programming, greedy algorithms, divide and conquer) and analysis methodologies.

Foundations of Artificial Intelligence

MS level at PPGC-UFPel (typically on first semesters). This course provides an overview of the main concepts and techniques behind Artificial Intelligence. Exhaustive and heuristic searches, knowledge representation, multi-agent systems, evolutionary computation, learning. Students are required to have extensive knowledge of some popular programming language, algorithms and data structures, as this course is very project-oriented.

Other activities

Programming Challenges

Programming Challenges are contests where teams of students compete to solve several challenges by programming. I have been coaching and helping organize regional sites for Maratona de Programação, part of the ACM ICPC, for over 6 years.

More about this
Maratona de Programação