Essay 1 Marta Martinez-Camara Feb. 2014

Imagine that as an Anita Borg scholar, you are given the opportunity to give a 5 minute speech to a group of high school students to encourage them to pursue degrees in Computer Science. Remember, the students want to know what is exciting and interesting about the field. Please prepare a speech to give to the students. (300-500 words)

I was in your situation not so many years ago. I remember how confused I was about my future, about what to do with all those opportunities. And at the same time, I also remember how excited I was, and determined to do something big, something truly rewarding with my career. I decided to follow the path of computer science without knowing very well why. It was just my intuition which was telling me that I would not get bored there, and that I would really see meaningful results come from my work. Nine years later, here I am, pursuing a PhD in CS. And one thing is for sure: my intuition was not wrong. No other field has the capacity to change the world like CS can - and does! Think about it: government, education, medicine, economy, visual arts, music, and the list goes on. They have all been deeply transformed in the last twenty years by the developments in CS. More importantly, it is quite clear that we have only seen the beginning of this transformation. Hands-on participation in this change, and the feeling of having a front row seat on our journey into the future: this is excitement that few other professions can give you. This journey is reflected in my everyday work: constantly learning new concepts and the need to be extremely creative, in order to deal with all the problems that pop up along the way. The most amazing thing though, is that I am always surrounded by people who are continuously learning and are extremely creative.a Just in case you were wondering, I would like to tell you one thing: not everybody who works in CS is just coding. You need to know how to code, because this is a tool, one that is used in almost every other technical job - mathematics, physics, economy, marketing, etc. But in fact, in CS, you will spend almost all of your time learning new things and solving cool problems. I said cool problems because they really are! Let me give you some examples: to make more realistic textures in animated films; to capture and reproduce realistic images of art pieces; to determine how much radioactivity was released in the Fukushima accident; to understand how a disease is spread in a population; to develop non-invasive techniques to detect breast cancer; to create free online education for everybody. Cool, isn’t it? When I started my PhD work, someone asked me what I would like to achieve. I replied that I would like to contribute to the creation of a better world, and a better society. A few years later, I am still convinced that I am in one of the best places to do so.

Essay 2 Marta Martinez-Camara Feb. 2014

Please write an essay on a technical project you took part in, or on a piece of research you undertook, where your contribution and involvement was key to its success. When writing your essay, please remember that the CS professional reviewing your application may not share the same technical expertise or knowledge of your particular research field. Please make sure to explain all technical terms and processes accordingly. Your essay should include the following sections: The problem your project or research is trying to solve, the solution that was chosen, the technical challenges you faced, your contribution to the success of the project and why you consider this project successful or innovative or both. (400-600 words) Note: Treat this essay as a technical report or research paper. Feel free to use tables, references, or figures.

Pollution, and in particular emissions of harmful substances into the atmosphere, is one of our main environmental problems nowadays. Laws are being developed by international organisations to control such emissions, but a technical framework is still missing to estimate the emissions correctly from a distributed set of sensors around the globe. Besides this, governmental agencies often deny access to the necessary measurements to the research community. This makes the development of new estimation methods much more difficult. One particular case of harmful emissions took place during the Fukushima accident, when unknown quantities of radioactive material were released into the atmosphere. The japanese government provided little information about the emissions, and this information was often not reliable. Thus, worried japanese citizens decided to organize themselves and collect measurements with mobile sensors attached to their smartphones. In this way, they created a public-access dataset of more than 2 million measurements. However, little can be done with these measurements if we do not have appropriate methods to estimate the releases. The project I worked on focused on developing such estimation methods, and in particular methods tailored to the Fukushima accident. Atmospheric dispersion models describe the trajectory of the released particles, relating emissions and measurements in a linear way: y = Ax, where y are the measurements, A is the dispersion model and x the ”source”, i.e., the quantity of material emitted over time. In this case, y and an approximation to A are known, and x should be estimated. Hence, to estimate x we must solve a linear inverse problem. My first contribution to the project was to formulate our problem in this framework, using the radioactivity measurement and the weather data that we had available. There are two reasons why this linear inverse problem is especially hard. First, the approximate dispersion model A is far from being perfect – it contains errors. Also, the condition number of A is typically very large. The simplest way of solving a liner inverse problem is using the least-squares method ˆ = arg min||y x x

Ax||22

(1)

ˆ is extremely sensitive to errors. To avoid this, we should introduce more But when A has a large condition number, x information about the problem in the objective function (1). In this case, I decided to use the fact that the radioactive materials were released by explosions, which means that the source x should be ”sparse”. This is why I added the `1 norm to the objective function, which had never been done before for this problem. Also, for physical reasons, the source x must be non-negative. I added this constraint too, thus resulting in the following new objective function: ˆ = arg min||y x x

Ax||22 + ||x||1 s.t. x ⌫ 0.

(2)

The estimation obtained with this method matches with the main events during the Fukushima accident (see Figure 1).

Estimated Releases of Xenon at Different Altitudes [105 GBq/s] 3

1000 m

3

3 300 m

50 m 0m

2.25 1.5 0.75

2.25 1.5 0.75

2.25 1.5 0.75 11th March

12th March

13th March

14th March

15th March

plo

Un

de

it 3

s*

*

* -U 3 U1

21

:00

U4

Ex

Ve nti ng

20

:20

Ve nti ng :00 ~4

Ea r th qu 21 ak :30 e Ve nt ing U1 *

UTC time

Figure 1: Estimated source. Hence, this new method was both innovative and successful. Previously the source was estimated with methods which use an a priori ”guess”. The introduction of this a priori information strongly biases the solution of the inverse problem (see [1]). This is why it was important to me to come up with a method which avoids such a bias. There is still a long way to go until we are ready to apply these method automatically to crowdsensing datasets. Atmospheric dispersion models – which estimate A – need as input terabytes of publicly-available high-quality meteorological measurements, and take days of processing to generate a single model. Meteorologist are working hard to simplify these models and to make them applicable to every kind of measurement dataset. This together with further development of our source estimation methods takes us a step closer to putting helpful information in the hands of the citizens themselves.

References [1] M. Martinez-Camara, I. Dokmanic, J. Ranieri, R. Scheibler, M. Vetterli, and A. Stohl, “The Fukushima Inverse Problem,” in 38th International Conference on Acoustics, Speech, and Signal Processing, 2013.

2

Essay 3 Marta Martinez-Camara Feb. 2014

Please provide 1-2 examples of a time when you exhibited leadership. Explain what you were trying to achieve and how you were influential. Feel free to examine the ways in which you acted as a role model to the members of your broader community, your technical community, or your university. (400-600 words)

In the first year of my PhD, I started working on a project with two other colleagues, as well as my advisor. Soon, we arrived at an impasse: we needed measurements of radioactivity and weather matrices associated to these measurements. We did not have this data, and no progress could be made without it. The weather matrices had to be generated by running simulations on very specialized software, implemented in FORTRAN by meteorologists. The software needs as input terabytes of weather information, the simulations to generate one single matrix take days, and the software is far from being user friendly. On the other hand, the measurements of radioactivity are highly confidential; access to them requires special permits given by international organizations. Just installing the software for generating the matrices took me two weeks. After hours of phone calls I learnt that obtaining the permit to access the radioactive data will take me around one year, lots of paperwork, and the whole process was not guaranteed to end in success. Meanwhile, I read a paper by a group of Norwegian researchers where these weather matrices were used, as well as the radioactive data. It struck me that the know-how of these researchers was clearly on creating the matrices and gathering the data, but the significant improvements that could still be made on the later processing of this information was not their expertise - but it certainly was mine. I decided to take a lead in this situation. My proposal to the team was to establish a collaboration with the Norwegian group, where each team would bring in their expertise. This would allow us to set completely new and ambitious goals, that were beyond the reach of either group in isolation. I established contact and a week later, I was in Norway for a one-week stay; a month later we had our first results; five months later we had our first prestigious publication together. Today, our unique collaboration continues strong, keeps bearing significant fruit, and forms the backbone of my PhD work. I am especially proud that more researchers from other universities are now expressing their interest in getting involved in this collaboration that I initiated.

Essay 4 Marta Martinez-Camara Feb. 2014 Dr. Anita Borg proposed the 50/50 by 2020 initiative, so that women earning computing degrees would be 50% of the graduates by year 2020. The percentage of Computer Science degrees earned by women, however, is still far from 50% in most countries. Please write a 600-1000 word essay which includes the following sections: Based on your experience/observation at your university, what percentage of women study Computer Science? What is your university doing to encourage women to select technical degrees? What cultural factors in your country influence fewer women to select technical degrees? If you were the Head of the Computer Science department at your university, what initiatives would you start to reverse the trend and increase the involvement of women in Computer Science?

Let me be frank: I wish I were not applying for a women-only scholarship. I wish the question of gender in CS were o↵ the table, and pure merit and ideas is all that were discussed here. But just looking around me, I see that unfortunately scholarships like this one are still necessary. There are few girls at my school at EPFL - just 10 to maybe 15%. One tries not to let this a↵ect her, but being in such a small minority is sometimes hard, and always so blatantly evident. The situation is not better when I look around farther. Just last week I visited the Zurich offices of a major global CS company - a true giant that seems to be able to achieve anything - except gender balance. At all the international conferences in my field, the situation is the same. Surrounded by so many smart people, I can’t help but wonder: where have all the girls gone? I recognize that there are many complex reasons that brought about the gender imbalance in CS in Spain and Switzerland. As I understand, biases start very early on. Deeply ingrained cultural beliefs of what is ”fit for boys” or ”fit for girls” are reflected in the choices of toys given to young children. Unwittingly, parents start us o↵ on well-worn paths that have brought us to this situation in the first place. In adolescence, social aspects kick in. In these years, opinions are formed from vague impressions and popular beliefs, often with little fact and much peer pressure. Teenage years are where the ”fit for boys” and ”fit for girls” prejudices finally get to influence the choice of future career - again often without much real information. Once this teen barrier is passed, there is no turning back. Like in many other places, specific programs to fight the gender imbalance also exist in Switzerland. EPFL has a collection of interesting programs. They are mainly focused on young children (7-13), young graduates (potential PhD students), and scientists (current PhD students, postdocs, and sta↵). In my opinion, this mix misses the main point: the crucial significance of the teenage years. I would propose a 3-pronged approach. First, actively change the misperception of a career in CS by teen girls. The distorted idea of our profession is rooted in a real lack of information. Hence, courses of introduction to CS should be instated, where not already available. Along with the core technical concepts of CS, students should also experience what a CS career actually feels like. Guest lectures by practicing scientists and engineers would be a regular fixture. Second, make the female leaders in CS more visible. This ties into the first point strongly. Role models serve to dispel false (but popular) beliefs and provide a strong incentive to pursue a similar path. This point is also one of the aims of the Anita Borg Memorial Scholarship. Third, foment the development of technical skills in young girls. To break the prejudices of what toys are ”fit for girls” is to address the root of the gender imbalance problem. Hence, I would create a new line of toys specifically designed for future girl engineers (I was very glad last Christmas to see a toy like this!). They would help develop all the necessary skills, and still be a lot of fun. It is precisely this enjoyment that will later on grow into a realization that pursuing a technical career can be fun and exciting!