TU Delft focuses on AI
TU Delft is intensifying its research into artificial intelligence, doubling its budget to €70 million per year, with 24 interdisciplinary AI labs.
AI has really taken off, playing a role in almost all scientific fields. That means it’s time to really focus on it. Eight new ‘TU Delft AI Labs’ have been set up at the university and that figure is set to increase to 24 by 2021.
The research will start with eight research groups in fields including material science, urban environment, water management, computer vision, biomedical applications and human-computer interaction.
John Schmitz, Dean of the Faculty of Electrical Engineering, Mathematics and Computer Science, is one of the driving forces behind the upscaling of AI research. In his view, the great thing about the labs is that they are focusing on fundamental issues as well as applications. “We plan to tackle issues that are of relevance to society.”
“Especially in medicine, I expect AI to bring about major advances. Take Alzheimer’s, a disease linked to countless genes. That means you’re dealing with an awful lot of data and it’s difficult to gain an overview. Self-learning algorithms could provide an answer. The Bioinformatics group is already researching this.”
There are numerous other areas where Delft researchers could make a difference with AI. A selection of the projects:
Water management
Climate change is making urban water management increasingly complex. There is a fundamental tension between supplying sufficient drinking water and limiting flood risks. Fortunately, there is plenty of data. There is a lot of digitisation in the water sector, which means that self-learning algorithms can help in devising sustainable solutions.
In the <link ai ai-labs aidrolab>AidroLab, TU Delft scientists aim to use geometric deep learning technology. This takes account of the complex mutual relationships of water networks and urban systems.
Happy in the city
How happy are people in their cities? The statistics traditionally used to predict the impact of policy interventions in cities provide only a limited picture, covering such areas as population density, property values and income levels. Values such as attractiveness, safety, liveability and accessibility often remain hidden. Although these values may be less tangible, they often matter more for people. This is why AI is being developed in the CTAI Lab to use big data, such as streetview data and Twitter feeds, to improve understanding of quality of life in cities.
3D city for self-driving cars
Colleagues in the <link ai ai-labs>3D Urban Understanding Lab are also focusing on cities, developing self-learning algorithms that can chart cities in 3D. The results are very detailed, showing every lamppost and road sign. The models being developed in Delft could eventually prove useful for self-driving intelligent vehicles. They can also play a role in simulations to evaluate the effect of various scenarios, such as floods, traffic, energy, wind, pollution and noise.
Computer says no
Humans are superior when it comes to devising creative solutions in unexpected situations. AI can make logical decisions based on mountains of data. This means that humans and machines can complement each other. But according to the researchers in the <link ai ai-labs aiman-lab>Transparent & Traceable AI in Human-AI Teamwork lab, it won’t happen automatically. This lab aims to develop instruments and methodologies to enable humans and AI to understand and supplement each other.)