Developing New Research Tools

The Laboratory of Algorithms and Technologies for Networks Analysis (LATNA)  in Nizhny Novgorod is one of HSE’s international laboratories. It is co-supervised by Panos Pardalos, Distinguished Professor at the University of Florida, and Valery Kalyagin, Professor and Department Head at the Department of Applied Mathematics and Informatics. Professor Kalyagin told The HSE Look about the lab’s research projects, its work with students, and engagement in the educational life of the campus and HSE at large.

What is the laboratory’s main research focus? What kinds of issues or thematic areas can network analysis be applied to?

Network analysis is quite a wide field of research, and HSE has developed several laboratories, which can study its different aspects. For instance, our laboratory  LATNA focuses on algorithms, i.e. on developing the tools for network analysis, while our colleagues at International Laboratory for Applied Network Research focus closely on applications, in particular to social networks.

One of the important research areas regarding networks analysis is optimization on networks. In terms of theoretical research, we analyze computational complexity of different optimization problems on networks, which shows whether it is possible to get exact solution of the problem in a practical amount of time. Efficient algorithms give a solution close to the optimal for such tasks, and we use the theoretical advancement to develop efficient algorithms for addressing practical issues.

 
One of our projects is devoted to clustering and search techniques in large-scale networks, where we focus on efficient algorithms to find closely connected communities within the network using different criteria. It’s applicable to a wide array of tasks, including analysis of social networks, gene expression networks, market networks and many others.
 
Another thing worth mentioning is our work on search algorithms, which has a particular focus on finding specific objects quickly in large databases. We rely on network analysis to make searches faster and more reliable, for example, in regards to image or face recognition.
 
Are you engaged in many applied research projects? Who are the lab’s major partners?
 
We have received a fair amount of such requests. Of course, not all of them turn into collaborations, but we’ve had several successful projects in logistics, optimization and data analysis. For example, a leading retail company needed to develop more optimal transportation routes. So, our researchers not only offered a practical solution for this client, but also found a way to develop a theoretical understanding of the tools used, and are currently working on that.
 
We’ve worked with INTEL Company on testing their Intel Data Analytics Acceleration Library and developing a series of laboratory works for students in applied mathematics. Recently we’ve started collaboration with Medical University together with our colleagues in cognitive sciences.
 It is quite a monumental and cross-discipline task to develop rehabilitation devices, which can treat speech impairments or poor coordination. This project is going to be a blend of medical research, cognitive sciences, psychology and data analysis.
 
Professor Pardalos and I do our best to encourage young researchers in our laboratory to more boldly engage into applied projects, because their level of theoretical understanding allows to go deeper inside the problem and to use the latest advancements for practical application.
 
You hold regular seminars. What is their goal? Are they open to lab members only or can anyone join?

Our seminars are a way for all lab members to discuss latest research developments and projects together. At the same time, it is open to everyone, and the schedule is available online, so that colleagues from other universities in Nizhny Novgorod can join us. It is a great way for young researchers to learn and grow as scientists, and we invite distinguished speakers when we get the opportunity. For instance, when Maurice Salles, Professor at Université de Caen Normandie (France), visited the April Conference earlier this year, he also gave a talk at HSE Nizhny Novgorod.

As a part of the laboratory’s seminars, we also participate via videoconference in a seminar ‘Mathematical methods of decision analysis in Economics, business and politics’ organized by Professor  Fuad Aleskerov at the Department  of Mathematics on the Faculty of Economics HSE Moscow, as well as in the seminar for the PhD students of the Computer Science Faculty. It’s a great way for our researchers not only to listen to the proceedings, but also to ask questions and participate in the discussion despite the physical distance between Moscow and Nizhny Novgorod.

How do you involve students in research?

Around 70% of our laboratory is under 35 years of age, including two undergraduate students, two Master’s and five PhD students. It is our strong belief that involving students in research is imperative. Prof. Pardalos pays a special attention to the support of young researchers; he is one of the most prolific academic supervisors in USA, with more than 50 successful PhD students.

We are closely involved in the Bachelor’s programme in Applied Mathematics and Informatics, as well as the Master’s programme in Data Mining. Many of our researchers teach courses for BA and MA students, so they can offer the most recent knowledge and methodologies in the classroom, as well as encourage students’ interest in research opportunities. Every year, we get around 20 students in total, who actively take part in the lab’s projects.

In addition, we organize two schools for young researchers, both from HSE and other universities. This is the Summer School on Operational Research and Applications (since 2012)  and Winter School on Data Analytics (since 2014). Students have a chance to meet with world-class experts in various fields of computer science and data analysis. This allows us to identify students with the potential for doing research, and they, in turn, can get familiar with the latest approaches to data analysis, as well as the current research activities being carried out at LATNA.

What other HSE labs and departments do you collaborate with?

As a part of the Faculty of Informatics, Mathematics and Computer Science, we work closely with its other departments. What is perhaps less typical, but no less important for us, is our collaboration with Fundamental and Applied Linguistics Department, which began from a desire to find a common ground for researchers in humanities and IT to work together at HSE Nizhny Novgorod. We have a strong profile in natural language processing, both in theoretical research and offering practical solutions to some of the city’s companies engaged in text analysis services.

If we talk about HSE at large, I’ve already mentioned that we collaborate with the Faculty of Computer Science, with the Department of Mathematics at the Faculty of Economics in Moscow and with the Laboratory of Applied Networks Analysis - we participate in each other’s conferences and organize joint sessions, exchanging the latest algorithms and discussing research problems. This process is very fruitful, as our research areas do not overlap 100% but are nonetheless complementary.

Professor Pardalos encourages the lab staff to actively participate in the development of international visibility of the lab and its reputation among the professionals in Computer Science and Data Analysis, and we continue to focus on developing the fundamental research, with its strengths, but also to always look for new ways to use it for solving practical problems.

Read more in The HSE Look July 2018 issue