Gå til hovedinnhold
0
Jump to main content

He finds hidden connections with artificial intelligence

Kevin Roy received a best paper award for his work presented at the IEEE International Seminar on Machine Learning, Optimization, and Data Science (ISMODE) held in Jakarta, Indonesia in 2022.

Kevin Roy has developed an algorithm that can detect causal relationships in complex systems where several sensors are involved, and where several tasks are performed simultaneously.
Kevin Roy has developed an algorithm that can detect causal relationships in complex systems where several sensors are involved, and where several tasks are performed simultaneously.

The work was selected among papers from authors affiliated to Google, Adobe USA, Carnegie Mellon University and other top companies and universities from more than 20 different countries. 

“The paper is a part of my Ph.D. thesis, which I plan to finish and deliver at the end of this year,” says Kevin Roy, a Ph.D. Research Fellow at UiA´s WISENET Center.

AI discovers hidden connections

Kevin Roy has developed an algorithm that is able to discover causal relationships among different parts of a complex system. 

“The algorithm analyses the signals captured by a network of sensors monitoring an industrial process,” he says.  

The advantage of an AI system that discovers hidden connections is that the offshore industry can improve and streamline its systems.

“To simplify, you can say that the sensors monitoring temperature, water level, pressure and other things might co-operate with each other in a way humans cannot interpret but the algorithm can. For instance if a sensor brakes down, we are able to go back to the algorithm and see how it is connected with other sensors and why it broke down,” he says.

Kevin Roy illustrates what kind of equations are needed to create an algorithm that can learn to interpret and see relationships in complex data and network systems on its own.

Kevin Roy illustrates what kind of equations are needed to create an algorithm that can learn to interpret and see relationships in complex data and network systems on its own.

A possible tool for several sectors

The algorithm was proven to be successful when tested with signals from sensors monitoring an industrial process, but Roy points out that its use can be extended to other fields as well.

“The algorithm can be used in any field where there is a need to understand large-scale dynamic systems, such as the financial and health sectors,” he says. 

A significant contribution from this research is that the algorithm has improved accuracy, and demands less computational resources and data as compared to what you would normally need when you use AI.  

“In addition, our method is easily explained, which traditional AI methods like deep learning are not,” he says.

Creating knowledge together

The research is part of the SFI Offshore Mechatronics project funded by the Research Council of Norway. In this project, Roy and his collegues at WISENET work with Data Analytics, IT Integration and Big Data.

Roy emphasises that the prize winning paper has been created in collaboration with his supervisor Professor Baltasar Enrique Beferull-Lozano, co-supervisor Luis Miguel Lopez Ramos and other colleagues at UiA´s WISENET.

Roy and Beferull-Lozano are both happy for the contributions from Lundin Energy, and MHWirth Norway (now HMH) for providing them with data from the oil and gas plant to carry out this research.