Real-life data ask for strong algorithms: Mitra Baratchi designs them
How do we deal with large sources of greenhouse gases? Do schools provide a socially-inclusive environment for all children? And how can we protect Earth’s nature? These questions have two things in common: they are complex global challenges, and data can help answer them. Mitra Baratchi is computer scientist and designs algorithms to better use data. She has received the Aspasia premium for her work with so-called spatio-temporal data.
Everything around us happens at a place and a time. When you can capture the value of a certain parameter with a sensor, you can collect spatio-temporal data. That is data with a place (spatial) and a time (temporal) component. Traffic sensors, mobile phones, satellites and even telescopes: these are all sensors that are measuring something over a large space and over a long period of time. Sensors can measure many things, from methane concentrations to movement of children playing at school yards.
Real-life data are far from perfect
These data can help us understand what is going on in the world. Much of these data do not reach its full potential, because it needs some complex processing. As a computer scientist, Mitra Baratchi designs algorithms that can make better use of data. While processing these data, you run into all sorts of problems, she explains.
Mitra Baratchi designs algorithms to better handle complex and imperfect data.
For instance, some data could be missing, because the sensor failed. Or you want to combine data sources with different qualities or resolutions. Real-life data are far from perfect. But these are the problems we have to deal with. Baratchi: ‘I am trying to automatically handle all these challenges. We want algorithms to easily generate models for similar datasets rather than asking users to design a model for each dataset separately. That way, users can easily utilise all these data sources.
Data and knowledge are two different things
Who are these users? Baratchi started working more and more with other scientists that collect spatial-temporal data. They have a wealth of data to address problems, but not the expertise to make the best possible use of complex and imperfect data. Baratchi, on the other hand, realised that the detailed domain knowledge these scientists possess is essential for understanding and solving many of the problems. She has worked extensively with environmental scientists and, more recently, astronomers.
How can data help in making decisions?
Baratchi is interested in complex global challenges, such as climate change, and how data can help in informed decision-making. Existing algorithms often allow us to predict what happens over time based on the assumption that things are constant and do not change unexpectedly. ‘But what we’re trying to do is different. We want to be able to predict how potential interventions might impact other processes in a complex, interconnected world. This calls for algorithms that allow us to acquire a deeper level of understanding.’
‘Data can play an important role in our decision-making process’
Machines cannot decide for us
And this deeper level of understanding is very challenging to reach with only data, Baratchi realises, because data-driven knowledge often lacks other scientific knowledge. That’s why there is not some future scenario in which machines reliably make decisions for us, according to Baratchi. ‘There is always a level of uncertainty with prediction models.’ Still, data can play an even more important role in our decision-making process than is currently the case, says Baratchi. ‘It can help us better understand what’s going on, and what’s going to happen.’
About the Aspasia premium
The Aspasia premium is meant for female scientists applying for a Vidi grant. If the applying scientists are deemed very good or excellent, but do not receive a Vidi, they can get the Aspasia premium. This premium is awarded under the condition of the female scientist being promoted to associate professor and amounts to € 120,000.