Tallinn University of Technology

TalTech actively cooperates with companies to ensure that theoretical research would immediately find practical outlets. A good example from recent times is the cooperation with the technology company R8 Technologies OÜ.

Molika Meas magister TalTech
Molika Meas

Author: Ketlin Rauk

Juri Belikov is a Professor at the Department of Software Science and the Head of the Nonlinear Control Systems Group of TalTech. He supervised the master’s thesis of Molika Meas, who interned at R8 Technologies OÜ. She later took up a job in the same company and conducted her research* at the same time.

In the interview, Juri Belikov, Ahmet Köse (the co-founder and Head of the Development Team of R8 Technologies), and Eduard Petlenkov (Professor at the Department of Computer Systems and the Head of the Centre for Intelligent Systems of TalTech) explain the true value of the outstanding thesis and the cooperation that has started between the university and the company.

Please tell us more about the master’s thesis of Molika Meas – what would you consider its most important outcome?

Juri Belikov: ‘As a result of the thesis, we developed a novel methodology and algorithms and published them at an international conference and in a scientific article. We worked with the company to develop a solution that helped take their product to the next level. The thesis was also awarded a special prize at the IT Innovation Festival of TalTech.’

How exactly did this project help the company?

Ahmet Köse: ‘R8 Technologies is an Estonian start-up that has developed a unique software tool – the R8 Digital Operator. It uses artificial intelligence to control commercial buildings remotely, ensuring the best indoor climate in buildings at minimum cost. The R8 platform collects data in real time (24/7) and learns the dynamics of a building to optimise its energy performance. Artificial intelligence decides the steps for maintaining the indoor climate and keeping costs to a minimum. However, it can be difficult for customers to trust AI – will it make the right decisions? What is the basis for its decisions? The new EU guidelines that will soon come into force require AI-based solutions to be fully transparent and easy to explain. So, it was important to create a trustworthy artificial intelligence whose decisions would leave no room for doubt. That is where the idea for the collaboration between the company and the university came from – to create a solution for artificial intelligence to be able to explain and justify its decisions.’

Can you specify what the most important thing about teaching an AI is?

Eduard Petlenkov: ‘The AI understanding things correctly is important. Let me give you an example: we are teaching schoolchildren so we give them a multiple-choice test (A, B, C, D), whereas the correct answer is usually A. A child’s mind learns in exactly the same way as artificial intelligence. They learn quickly how to give the ‘correct’ answer, and even if it is right, can we trust it? You have to ask ‘Why?’. This is precisely the question that explainable artificial intelligence answers. Therefore, it is always important to know why artificial intelligence decides in favour of a certain thing.’

What has changed for the customers of the company as a result?

Ahmet Köse: ‘The updated system is simpler and more transparent. A person maintaining a building will receive information from the device along with explanations. This way, they can immediately see where the problem is and what is causing it, and can make decisions accordingly on how to solve the problem, who to call in, etc. For one thing, this information makes the system more reliable for the customer; it also helps save time and money because you no longer have to search for the problem – it will be immediately obvious.’

Why is it important for a university to work with businesses?

Juri Belikov: ‘First of all, because we can see immediately that the work done at the university does not stay on paper but is actually of interest to the industry. We can proudly say that the theoretical research in this example was implemented in practice straight away.’

Eduard Petlenkov: ‘Putting knowledge into practice and starting to apply it is a huge load of work, a science in itself! We are grateful to R8 that our agreement allows us to publish the results of our collaboration and make them available to the scientific community.’

Ahmet Köse: ‘Our cooperation was extremely successful and it will definitely continue. We will also continue our research – exploring, researching, developing, and trying to help people around the world with our IT solutions.’

The master’s thesis of Molika Meas ‘XAI-based fault detection, diagnostics and monitoring method for air handling unit’ is available here.