André Martins Gets a Ph.D. and is Hired by Priberam
|On May 11, 2012, André Torres Martins became the first dual degree student to finish a Ph.D. in Language Technology (LT), at Instituto Superior Técnico of the Universidade Técnica de Lisboa (IST/UTL) and Carnegie Mellon University (CMU). he became the fourth student to accomplish this goal in the scope of the Carnegie Mellon Portugal Program. Alexandre Mateus (Engineering and Public Policy) was the first one, then Rita Ferreira (Applied Mathematic) followed, and at the end of 2011 Ana Venâncio (Technological Change and Entrepreneurship) finished her Ph.D. All of them finished their Ph.D.’s in different programs.|
In a little interview, André Torres Martins speaks about his future and also about the impact of his thesis titled “Advances in Structured Prediction for Natural Language Processing,” (http://www.cs.cmu.edu/~afm/Home_files/thesis.pdf ) carried out with his four advisors: Mário Figueiredo (IST/UTL), Noah Smith (CMU), Pedro Aguiar (IST/UTL), and Eric Xing (CMU).
CMU Portugal: You finished in May your dual degree PhD. In looking back, what were your main challenges?
André Martins (AM): The whole experience was very challenging. My early Ph.D. years were spent at CMU, where besides doing all the coursework, I learned about all the challenging problems in natural language processing and machine learning. Given that I had so much to learn, it was challenging to make the most of those two years, but overall it went very well. It wouldn’t be so without the help of all my friends and family during our stay in Pittsburgh. Research-wise, I had to deal with four advisors – which multiply by four the number of challenging problems! – but I was very fortunate to have the support of all of them. Perhaps the most difficult challenge was to put all the material together to write a coherent thesis.
CMU Portugal: What kind of impact does your thesis has to the area of Language Technology?
AM: My thesis paves the way for developing more accurate statistical models of language, which will have an impact in search quality, translation systems, and text analytics. While most previous work resorts to oversimplified models for the sake of computational tractability, I propose a new framework based on linear relaxations that incorporates rich linguistic knowledge without sacrificing efficiency. By using machine learning techniques, I show how the proposed framework yields efficient and state-of-the-art syntactic parsers for several languages.
CMU Portugal: What do you intend to do next?
AM: I will join the Machine Learning research team at Priberam, a spin-off company from Instituto Superior Técnico of the Universidade Técnica de Lisboa. Priberam has been developing language technology for more than two decades. With the advent of social media and the availability of large amounts of textual data, there are many opportunities to apply some of the machine learning techniques I learned during my Ph.D. to develop new products. There are lots of interesting research to be done.
Articles about André Martins:
Internship: “A Win-Win Process to the University, to the Company, and to the Student”
Portuguese Ph.D. Student Launches Priberam Machine Learning Lunch Seminars