Campinas, the new Silicon Valley ?

I have been bombarded so much by radiant news about Brazil that my stance about the country has slowly changed from cynically pessimistic to cautiously optimistic.

The latest attack on my scepticism comes from The Washington Post, and talks about the town of Campinas, where my University is located. The piece is not exactly new, but only came to my awareness today :

Brazil: Home of the next Mark Zuckerberg?

(N.B.: those bombastic titles don’t exactly help me to take the argument seriously.)

By , Published: March 21

Silicon Valley has led the world in innovation and entrepreneurship because of its culture of information sharing and mentoring. No other region in the world is like it. But things are changing. (…) One of the most impressive examples of this is in Campinas, Brazil—a small university town on the outskirts of Sao Paulo.

In June 2010, ten startups at the Softex incubator at the Universidade Estadual de Campinas decided to break free from the university incubator they were housed in and form an entrepreneurial co-op of sorts, called the Associação Campinas Startups. Instead of relying on local business executives and professors to guide them, the entrepreneurs decided to learn from each other. (Read the rest)

I am, of course, delighted by the news, and I am sure the people both at Softex and at the new Associação Campinas Startups are doing a wonderful job. Last year, I’ve got acquainted with professionals and students involved in Campinas-area entrepreneurship projects, and I was very impressed by their enthusiasm and commitment. The administrative environment, however, has to help, or at very least, it has to avoid disrupting. I would be more willing to believe in our chances in the global innovation competition if it weren’t for our Kafkaesque bureaucracy and titanic tax burdens.

Wadhwa ends his piece with the optimistic prediction that “by the end of this decade, we will see some Mark Zuckerbergs emerging from the slums of Sao Paulo or New Delhi, India or Valparaiso, Chile”. My guess is that he’s yet to realize the depth of the inequality in Brazilian society. I don’t think it’s impossible that the next Zuckerberg will come from the favelas, but the chic neighborhood of Jardins in São Paulo has much better odds. The potential favela Zuckerberg may count herself lucky if she learns enough math to properly understand a cross-multiplication of proportions. By the end of high-school (in the already unlikely scenario she completes it), her chances of actually coping with high-school level math and science are as good as those of winning the lottery.

Monteiro Lobato Doodle

I’ve coment a while ago on the Viscount of Corncob and his Algebra Congestion (of which I am also a constant sufferer). The Viscount, together with Emilia, the Marquise of Shortail were pictured in the Google Doodle of today’s Brazilian search page. Both are the creation of Brazilian writer Monteiro Lobato, whose 129th birthday is commemorated today.

Upcoming Talk: kNN Search and CBIR

I’ve been invited by Prof. Francisco Pelaez and Prof. Camila Barione of the Centre of Mathematics Computing and Cognition of the Federal University of ABC to give my talk on kNN Search and CBIR (Content Based Image Retrieval).

I will discuss my past work, showing the three methods I’ve proposed during my thesis on high-dimensional multimedia indexing on large databases. But I also discuss some of my new research pursuits, related to the use of very discriminant local descriptors, like SIFT, on complex semantic queries, which require generalisation.

The talk, in Portuguese, will be on Tuesday May 26, at the Block B, room A801 of the Federal University of the ABC, which is located at the Rua Santa Adélia, 166, Santo André — SP, Brazil, CEP 09210-170. Their phone number is +55 11 4996-3166.

If you have a Google Calendar, you can save the date by clicking below:

Upcoming Talk: Three New Methods for kNN Search

Prof. Ricardo Torres has invited me to the Institute of Computing of the State University of Campinas, where I am giving a talk on the work I’ve done on my thesis. I will explore the challenges of kNN search (also known as k nearest neighbours search, or simply similarity search) and discuss the three original methods I’ve proposed: the 3-way trees, which are based on the traditional KD-Tree with the addition of redundant overlapping nodes; the projection KD-Forests, my first attempt of using an index composed of multiple moderate-dimensional sub-indexes; and finally the Multicurves, an index based on the use of multiple moderate-dimensional space-filling curves, which has several nice properties like ease of implementation, dynamicity (tolerance to insertions and deletions without performance degradation) and avoidance of random accesses (thus making secondary-memory implementation easier).

The talk will be in Portuguese.

Back Home

Prof. Arnaldo Araújo invited me for a 4-month research internship in the NPDI lab (which stands, in Portuguese, as an acronym for Digital Image Processing Centre).

I’ll be working in cooperation with his students to explore how aggregates of local descriptors, in the form of bags-of-features or visual dictionaries, can help in the classification of video and image databases.

NPDI is one the first labs in which I’ve worked and I am glad to have this opportunity to work with them again.

Upcoming Talk: High-Dimensional Indexing and CBIR

I’m giving a talk on the workshop organised by the Digital Image Processing Centre — NPDI for the French-Brazilian project CAPES-COFECUB on Interactive and Content-Based Multimedia Information Analysis for Digital Video Applications.

My talk, entitled “Indexing High-Dimensional Data – Application to CBIR” explores my recent work on multimedia indexing, k nearest neighbours search (kNN search) and image identification.  Here’s the abstract:

“The troubles of multimedia information retrieval start at its most elementary operation: matching the high-dimensional feature vectors used to describe the data. In this talk, we will discuss how recent innovative methods are taming the infamous ‘curse of dimensionality’ and how they can be used in CBIR. The author will discuss his recent contributions to the advance of the state-of-art and his current research endeavours.”

My talk will be on Wednesday, April 8, at 14h30. It will take place at the UFMG Pampulha Campus, on the ICEx building, in room 2077. Registration to the workshop is free.

EDIT 14/4: The presentation, with narrative is available on my (brand new) talks and courses page.

Back to Brazil… and to an old research subject

The lack of recent updates has been result of the end of my Research Internship in France, and all the consequent rush in tying up loose ends there, and setting up new projects here. All this international coming back and forth (how much luggage can one reasonably accumulate in four months?!) will end up killing me.

A good piece of news, though: I’ve passed my qualification for maître de conférences (which basically means that the French government considers me apt to be a teacher/researcher).

* * *

I was in Rio de Janeiro for the Carnaval,  and I’ve had the opportunity to meet Dr. Eustáquio J. Reis from IPEA (the Brazilian Institute of Applied Economics Research), who is coordinating an interesting project to preserve and diffuse the Memory of Brazilian Statistics, using digitisation and the Web as the main tools.

Before I’ve got interested to the subject of high-dimensional indexing, the long term preservation of digital data (and its use to give access to cultural heritage assets) was my main research interest, as my M.Sc. dissertation reveals. So, it was with pleasure that I revisited the subject at an informal lunch, on the picturesque neighbourhood of Santa Tereza. We browsed important questions like file formats, standards for image acquisitions, search tools, the possibilities (and limitations) of character recognition, and the difficult balance between short term access and long term preservation every project has to achieve.

I am looking forward to see the new developments of this project, which concerns a type of cultural asset which is often overlooked, despite its enormous importance: quantitative data.

End of Winter Break

After the break, back-to-work under the snow in one of the harsher European winters seen of recent.

I am glad that I’ve got my seasonal flu before the break, since I have so much work to finish before coming back to Brazil in January 30. I intend to submit soon a journal article about multicurves (my high-dimensional indexing method based on space-filling curves), and I have some other results which I would like to submit to one of the conferences in the “first wave of deadlines” (ICDAR, January 22; ICIP, January 30; ACM KDD, February 2). And, of course, I have to accomplish my final role on the EROS 3D project, preparing the end-of-project report to be submitted later this semester.

Interestingly, an unusual lot of Brazilian colleagues have chosen to spend their Summer break in Europe (thus exchanging heat waves and tropical storms for snow and glacial winds), including two former teachers and two former classmates.  This unusual density of temporary “exilés” has been fortunate, in the sense that it gives the opportunity of many encounters, discussions, and… future cooperation!