Shaping synergies between science and industry
Human-computer interaction, machine vision, performance and energy scalability, mobile computing, and data center optimization – these are some of the most significant challenges that computer science will have to face in the years to come. In the Microsoft Research - Swiss Joint Research Center, a collaborative research engagement between Microsoft Research, EPFL and ETH Zurich, science and industry try to tackle these challenges together.
Q: ETH and EPFL have been collaborating with Microsoft for several years. How does ETH benefits from this partnership?
GA: Collaborations with industry in general and with Microsoft research in particular bring a very useful perspective on the IT world and how technology is evolving. By partnering with one of the companies driving such evolution, we are in a position to improve both our research and teaching. On the research side, we get a better understanding for user requirements, use cases, workloads, and for ideas that have been tried in practice and did not turn out to work as well as imagined. This shapes our research by enabling us to focus on medium and long term objectives grounded on today’s real needs and concerns. On the teaching side, this better understanding of existing systems flows into lectures and student projects, giving students a more accurate view of the state-of-the-art and of things to come.
In addition, the direct exchange with researchers a Microsoft is a valuable experience for master and doctoral students, be it through meetings and discussions here at ETH or through an internship at MSR.
Q: Why is the collaboration between ETH and Microsoft important to Switzerland both socially and economically?
GA: ETH has the task to educate the leading engineers Switzerland needs and to help maintaining the country’s competitiveness. Both tasks are intimately tied to our research activities. The collaboration with Microsoft – one of the few IT companies running a research laboratory that operates as such – helps our Department to focus and improve research activities. It also provides funding for doctoral students and increases the visibility of ETH activities and of its students at an international level. Further, it generates many ideas that eventually flow into the economy through start-ups, students joining companies where they apply the unique expertise gained through the collaboration and patents. Sometimes, these contributions are less quantifiable but not less relevant. An example for such a contribution is the increased perception of Zurich as an IT hub due to its unique combination of top Universities, local industry, regulatory framework, and excellent location.
Q: Which projects of the Swiss research do you find most relevant? Why is that? And what has been achieved in these projects so far?
GA: Well, I can speak mostly about the project where I am involved. We are exploring how to use reconfigurable hardware (FPGAs) for data processing. On an FPGA, it is possible to implement data processing algorithms that significantly differ from existing CPU based approaches due to the much higher levels of parallelism available and the possibility of exploring pipelined designs. These algorithms can then be used in hardware accelerators for systems processing large amounts of data and they also help computer architects to decide what functionality could be added to processors to better support data science workloads. So far, we have achieved a number of very interesting results in both algorithms as well as systems: we have shown that we can implement functionally complete data processing systems into microservers that require a fraction of the space and energy than conventional computers.
Q: In which areas of research can we expect future projects of the center? Why?
GA: MSR has a wide set of interest, much like our department. Future projects are likely to reflect existing trends in computer science: robotics and control, embedded systems, and large scale data processing and data centers; plus, of course, many topics related to these general areas such as distribution, security and privacy and algorithms.
Q: Switzerland has brought forth an increasing number of collaborations between industry and academia. Why do you think that is?
GA: Today in IT there are numerous highly complex problems. As companies focus on their immediate products, many questions arise about the future evolution of systems and the state of the technology in a few years’ time. The collaborations between ETH and industry creates a win-win situation where industry feedback and funding helps ETH develop new research directions that eventually establish the basis for the solutions companies will need in the medium and long term.