How to exploit very large data masses efficiently
Author
Prof. Erkki Oja
Member of the Steering Committee NRP 75 “Big Data”
Interview with Erkki Oja, data scientist and member of the Steering Committee, about NRP 75 and Big Data in general.
What were your expectations towards NRP 75? What would have been missing without NRP 75?
When NRP75 was under the planning stage in 2014, I saw it as a rather unique concept. There was a lot of talk about Big Data in those days, and NRP75 was one of the first research programmes to combine the two major aspects of this research field: the technological one, ranging from algorithms to demonstrations and applications, and the societal one, looking at the ethical, regulatory and legal aspects. The programme was also very timely: Big Data as a technology and concept was yet in its early stages, and the research programme had the opportunity to follow and in some cases lead the developments that took place during the next eight years. I think this is a good example of a public financing body anticipating and correctly forecasting the progress of a new field in the near future.
Can you explain to us what Big Data means to you?
My role in the Steering Committee was that of a data science researcher. Although I have been working on many applications over the years, I mostly look at Big Data as a technology – how to exploit very large data masses efficiently in any given real-world application to gain useful information. At the core of this lies a computational problem that is very demanding due to the volume, variety and quality of the data. It has been fascinating to see how novel computing architectures and algorithms emerge in the scientific community, including NRP 75, to tackle this problem and facilitate a growing number of useful applications.
Where do you expect the biggest impact of Big Data? What is nonsense?
From my point of view, that of a theoretical scientist, perhaps the biggest impact in the future may be the completely new way of constructing Artificial Intelligence methods, which greatly enhance the more conventional ways to process and analyse data in a multitude of application fields. AI is something that has been around for a long time, but it is quite recently that some highly innovative breakthroughs have been seen, notably in image, speech, and language processing. This new “Deep AI” is based on the idea that Big Data can be used as material for powerful new machine learning methods, especially deep neural networks. This yields impressive results even in applications that seemed far-away goals, such as natural language understanding and generation. This may have a big impact on automation all through the digital society. As for nonsense, one should be very careful about the hype: Big Data is just data without any inherent intelligence, and machine learning just reveals underlying connections and correlations in the data. It remains the task of the human user to make use of the technology in a critical and meaningful way.
How do you assess Switzerland’s position in terms of Big Data research and applications? NRP 75’s position?
Switzerland is in the front-line in the international Big Data research and applications. This is based on the traditionally very strong position in information sciences in general. There are excellent research groups in Swiss universities, and we managed to get many of them to participate in NRP 75. In spite of the limited resources of this kind of research programme, I believe NRP 75 gave a boost to this research. Especially the new connections and networks formed in this multi-disciplinary programme will certainly prove valuable. This contributes significantly to the innovation capacities of Switzerland and its attractiveness also for multinational companies and international organizations.