Alumnus Hugo Gonçalves talks about his dissertation and experience
in the Program
“Difficult Tasks Can Only be Accomplished if One
is Persistent Enough”
comes across as a hard-working researcher that is inspired in his work, and
does not shy away from a challenge. In July 2015, Hugo defended his thesis "Accelerated Sparse Coding with Overcomplete Dictionaries for Image Processing Applications,” which allowed him to accomplish his academic goals: to obtain a
dual degree Ph.D. in Electrical and Computer Engineering (ECE) at Universidade
do Porto and Carnegie Mellon University (CMU). Hugo’s dissertation was supervised
Miguel Velhote Correia, from Faculdade de Engenharia of the Universidade do
Porto, and Xin Li, from CMU.
The research focus of his Ph.D. was sparse
coding, a more mathematical topic than his original proposal, a change that put
him on the right track to the job he wanted. Hugo first started as an Analog
Design Engineer at Synopsys Inc., while he was still a dual degree doctoral
student, which he says was the perfect balance: “Working at Synopsys during the
day, and working on my Ph.D. at night and on the weekends made me do more with the
same time.” Synopsis Inc. is Silicon to Software™ partner for innovative
companies developing electronic products and software applications, which has
an office in Porto, Portugal.
CMU Portugal: Recently you defended a dissertation
titled "Accelerated Sparse Coding with Overcomplete Dictionaries for Image
Processing Applications.” What were your main findings?
Hugo Gonçalves (HG): Sparse coding
has many applications in several fields, but the algorithms used to solve the
sparse coding problem are considered time-consuming. The dissertation selected
the fastest algorithm through a benchmark among several algorithms, and
developed a few ways of accelerating the same algorithm. In particular, an
application of sparse coding for wafer probe testing, developed by my
advisor Xin Li, obtained a substantial reduction in processing time.
CMU Portugal: During your Ph.D. you had to change
your thesis proposal. Could you please comment on what you felt and the major
challenges you had to face at the time?
| Hugo Gonçalves with his colleagues at CMU.
|HG: When I started the CMU Portugal
program, I continued the research that was already planned. It consisted of
an ASIC (Application Specific Integrated Circuit) implementation of parallel
computing for sensor vision, which ended up providing unfavorable results.
Since the development cycle of an ASIC is very long, trying a different
implementation meant that I could not meet the dissertation deadline. This was
a big setback for my doctoral program, and so I was advised not to include it
in the dissertation. I had to take an alternative path. It was decided that I
would follow the algorithmic path, for which I already had a good basis, as I
was familiarized with sparse coding.
CMU Portugal: You published three papers with
your co-advisors in Portugal and at CMU. How do you comment on this experience
of having one advisor on each side of the Atlantic?
HG: There are pros and cons to having two advisors
in different locations. The pros are that I could benefit from two different
ways of thinking and different perspectives. On the other hand, there is always
one advisor who is far away and it is more difficult for him to be completely aware
of what is happening. Nevertheless, today we have several communication
channels that help us to communicate when necessary. The difference in time
zones was also an advantage and a disadvantage: on one side, the time overlap
was small, so it was more difficult to set up meetings, although my advisors
always managed to find a common time slot. Even still; this was also positive
because the period of the day that both advisors could respond was larger.
CMU Portugal: In looking back, what are your
fondest memories of this experience?
HG: I experienced new things mainly in Pittsburgh,
especially because I am from Porto and FEUP. I was amazed by CMU’s environment
and cultural diversity. Both inside and outside campus, things are very
different than in Portugal. Also, I had some great moments with the people I
met in Pittsburgh, especially my roommates. Of course, my stay in the United
States was a unique opportunity to visit cities that I would otherwise not have
the opportunity to visit soon. There were, however, some moments of adversity,
when nothing that I tried seemed to work. But in looking back, I feel that the challenges
I experienced during my research led me to never forget that difficult
tasks can only be accomplished if we are persistent.
CMU Portugal: You are now an Analog Design
Engineer at Synopsys Inc., in Porto. What led you to accept this position?
HG: As I said before, I designed an ASIC, which
was in fact made of analog components. Although I finished my dissertation with
a mathematical topic, I learned and practiced with a significant amount of
analog designs, which I enjoyed very much. SNPS Portugal was creating an analog
team in Porto, and looking for analog designers. It seemed a suitable position
for me: a highly specialized job in an international company, with challenging
problems, and doing something that I knew I would like to do. Regarding the
timeline, I accepted the position at a moment when I did not have a scholarship
anymore, and focusing uniquely on the Ph.D. was becoming counterproductive.
Working at Synopsys during the day, and working on my Ph.D at night and on the weekends
made me do more with the same time.
degree program was an opportunity to learn and practice the analog design that
I needed to get this position. Right now, I am sure that I made the right
choice to apply to this position and it would probably not happen if
I did not go through the doctoral program.” – Hugo Gonçalves