Finding the meaning in data
Calliope-Louisa Sotiropoulou, HLF15 participant. I first heard about the Heidelberg Laureate Forum from an email by the Association of Computing Machinery. I read all the available information and it immediately triggered my curiosity and interest. Such an opportunity to hear and interact with some of the most brilliant minds of our time and network with some of the most promising young minds in the fields of computer science and mathematics is not something to be missed. But immediately the question arose: as I am not a mathematician and not what someone would traditionally call a computer scientist (I specialize on digital hardware design), would I fit in? Some months later, here I am in beautiful Heidelberg wearing my name badge that states “computer science”, participating in this most exciting event. The interdisciplinary vision of HLF made all my “doubts” disappear.
In this brave new world of science we “young researchers” live in (by the way I love the definition of a “young researcher”. It feels like it has absolutely nothing to do with the age but absolutely everything to do with the mind) it is literally impossible for anyone to be strictly defined as belonging to a specific field of science. After all, interdisciplinarity is the key to any successful research, especially in applied sciences.
My academic journey can serve as an example. I started with physics, continued with electronic circuit design and ended up in embedded systems. In my research as a PhD student I have worked on application specific multiprocessing systems design and development. I specialized in image processing implementations on reconfigurable systems and had the opportunity to work on Biomedical Applications (machine vision) and applications for High Energy Physics (particle tracking for the ATLAS detector in CERN).
I am currently a post-doctoral researcher in the Department of Physics “Enrico Fermi” of the University of Pisa. There I continue developing hardware for the ATLAS detector in CERN as part of the Fast TracKer group, but my main research is on advanced image processing implementations for pattern matching and cognitive image processing.
The human brain is a real-time image processor continuously looking for contours in images to identify familiar shapes. The brain is believed to apply an early stage strong data reduction to identify features by using a pattern matching technique that identifies only those patterns that are “meaningful” for biologically plausible purposes (survival). This decision making process needs to be as fast as possible. The “meaningful” patterns change as the human brain trains and adapts to its surroundings.
The same principle is applied in particle tracking in High Energy Physics. In the Large Hadron Collider (LHC) high density bunches of protons are accelerated, running in opposite directions. These bunches cross each other at the center of each experiment built (ATLAS, CMS etc.) to record the proton-proton collisions with a frequency of 40 MHz (one crossing every 25 ns). The amount of data produced is enormous (1.9-2.0 MB / 25 ns ~ 80 TB/s). Therefore a quick pattern matching technique must be executed to identify particles with “meaningful” tracks and then store only the data from the collisions that are “interesting” to be studied and analyzed further.
What we see is that two scientific fields that would at first seem unrelated are being closely bound by the same problem: quick and efficient data reduction with minimized information loss. This is also what the “Big Data” problem is about (this year’s HLF Hot Topic): Fast decision making by processing huge amounts of data (quickly choosing “meaningful” data). The same or very similar problems are found in physics, biology, medicine, security, data centers etc. Solutions (both algorithms and hardware) that apply to one field can be adapted to be used in another, saving both time and effort. Therefore it is more “meaningful” to tackle such problems with an interdisciplinary perspective.
Being in this year’s Heidelberg Laureate Forum is a great privilege. HLF is a multidisciplinary think-tank. Scientists with varying backgrounds are here to learn, talk, interact, network. There remains only one challenge: make the most of this wonderful opportunity.
Calliope-Louisa is a post-doctoral researcher in the Department of Physics “Enrico Fermi” of the University of Pisa (also associated with the Istituto Nazionale di Fisica Nucleare – INFN). She is currently working on advanced pattern matching implementations for image processing of large data throughput applications. Such applications are particle tracking for High Energy Physics and cognitive image processing. Her research is funded by the Marie Curie FP7 IAPP Project: Fast TracKer for Hadron Colliders.