27 August 2024

On the 3rd of June, George Osipov at the Department of Computer and Information Science (IDA) successfully defended his thesis in the intersection of theoretical computer science and artificial intelligence (AI), with a focus on parameterized complexity of constraint satisfaction problems (CSPs).

image cut from thesis cover

Congratulations on your PhD! What was your background when you started your PhD?
"I have a bachelor's degree in computer science from Free University of Tbilisi in Georgia (where I'm from) and two years in a master's program in Mathematics at Ilia State University in Georgia with one Erasmus year in Uppsala."

What is it like to be a PhD student at IDA?
"My time as a PhD was quite pleasant overall. Despite having no prior research experience, close support from my supervisors helped me get up to speed quite quickly, and I have been enjoying research ever since. I was part of Wallenberg AI autonomous systems and software program (WASP) graduate school, which made the experience even more interesting with various courses around Sweden, summer and winter schools and international trips."

You have just finished your PhD. What do you want to do next?
"I am continuing to work on research directions that emerged in the later years of my PhD, now with the help of an International postdoc grant from the Swedish Research Council. It allows me to travel for two years outside Sweden, and my plan is to visit two universities in the UK: Oxford University and Royal Holloway University of London."

A summary of the thesis by George Osipov

Many real-world tasks can be phrased in terms of assigning values to variables subject to a set of constraints. This makes the Constraint Satisfaction Problem (CSP) a versatile modelling tool both in practice and in theory of computing. In applications, it is often impossible to satisfy all constraints, but there exist assignments that satisfy almost all constraints.

In his thesis, George Osipov has studied the complexity of solving almost satisfiable CSPs. Classical complexity theory strongly suggests that finding optimal assignments to CSPs is computationally intractable. However, by adopting the lens of parameterized complexity, many problems become tractable by leveraging the special structure of the instances. The thesis is at the intersection of theoretical computer science and AI and is not guided by an immediate application. One possible application is efficient data processing and automatic reasoning, which can be crucial for drawing insights from the drastically increasing amounts of data. An attractive feature of parameterized algorithms in this regard is that they scale favourablyto larger inputs.

To the thesis: On Infinite-Domain CSPs Parameterized by Solution Cost

Read more about George Osipov

Doctoral studies in computer and information science

More about WASP at IDA

Organisation

Latest news from LiU

Two men and a woman talk in front of a screen

Machine learning can give the climate a chance

Machine learning can help us discover new patterns and better tackle the climate crisis. Researchers from all over the world meet at Linköping University with the goal of finding and deepening collaborations in this area.

Sofia Nyström och Henrik Nordvall in front of a window that reflects the blue sky.

New course takes the UFO issue seriously

UFO – from tin foil hat to research topic. This is the name of a single-subject course that starts in the spring 2025 at LiU. It will approach the UFO issue in a serious way while looking at the question of what constitutes knowledge and truth.

Four female medical students enjoying the sun, study location Jönköping.

The medical programme at LiU wins award for student engagement

The medical programme at Linköping University (LiU) has won an international accreditation for student engagement and participation in shaping the programme.