Data science at the heart of six new interdisciplinary projects

Interdisciplinary research is often key to developing innovative digital solutions. Over the next three years, researchers from DTU Compute will collaborate with researchers from other academic fields at DTU and other universities on projects supported by the Villum Foundation.

Six DTU Compute projects have been awarded funding in the latest round of the Villum Synergy Programme .

All the projects combine data science with insights from other research fields to answer questions that no discipline can answer alone.

“DTU Compute’s involvement in half of the Villum Synergy grants awarded in 2026 reflects our strong position in data-driven and AI-based research, and the significant research potential that comes from working across research domains. We greatly value the opportunities provided by the Villum Foundation to develop new research collaborations across disciplines,” says Head of Department, Professor Jan Madsen.

Read more about the six projects below. If you want to contact the researchers, you will find their contact details at the end of the article.

PSYFIND: Psychological Fingerprints for Cyber-Deception with Typological Attacker Classification & Adaptive Defense.

Associate Professor Emmanouil Vasilomanolakis, DTU Compute – and Professor Ingo Zettler, Department of Psychology – University of Copenhagen. DKK 4.5 million.

Cyber attackers are no longer a single “type”: intrusions may be driven by automated malware, human adversaries, or emerging AI agents; still most cyber-security methods still treat them alike. PSYFIND will build personalised cyber defense by combining cyber-deception engineering with personality psychology to infer attacker type from interaction traces and adapt deception in real time.

The researchers will learn “psychological fingerprints” from behavior in deceptive environments using both large-scale honeypot deployments and controlled, ethically run ground-truth sessions.

The researchers will experimentally validate psychological mechanisms that reliably degrade human attacker performance and translate them into deployable, attacker-specific deception modules. PSYFIND delivers an integrated attacker-typing framework, open evaluation protocols, reusable software, and a curated benchmark dataset for next-generation security research.

“Cyber defence still largely treats every attacker in the same way. With PSYFIND, we want to change that. Our goal is to understand patterns in how different attackers behave, whether they are humans, malicious software or AI, and use that knowledge to build defences that adapt to the attacker rather than relying on a one-size-fits-all approach,” says Emmanouil Vasilomanolakis.

Physics-Driven Digital Twins for Two-Phase Cooling in AI Data Centers (SmartCool).

Professor Henrik Madsen, DTU Compute and Professor Jesper Hattel, DTU Construct. DKK 4.5 million.

he rapidly increasing energy density in AI industries is pushing for a major paradigm shift in cooling solutions from traditional air cooling to direct-to-chip liquid cooling.

Two-phase immersion cooling (IMC) can capture all the waste heat efficiently with a supreme heat transfer coefficient. However, the phase change process is associated with stochastic dynamics of microscale bubbles and a critical point for maximal heat flux.

SmartCool is aiming for controllable and predictable operations of IMC processes by establishing a framework connecting fundamental multiphysics models, advanced microfabrication of boiling surfaces, and stochastic greybox modeling.

The greybox models will improve our understanding of physical mechanisms of microscale heat transfer, at the same time, a squeeze-shift-adapt principle will be established for robust industrial implementation of IMC processes," Professor Henrik Madsen says.

The outcome of the project will pave the way for a broad spectrum of two-phase cooling solutions.

Data-driven design of tunable chalcogenide thin films for next-generation indoor and tandem photovoltaics.

Associate Professor Mikkel Schmidt, DTU Compute and Senior Researcher Stela Canulescu, DTU Electro. DKK 4.3 million.

To make solar energy more practical for everyday life, we need to move beyond large solar farms and also start powering indoor electronics, reducing our reliance on disposable batteries.

New materials like antimony sulfide show great promise for capturing indoor light, but they are currently difficult to perfect because even tiny mistakes during manufacturing can ruin their performance.

Instead of optimizing the design by old-fashioned trial and error, this project uses a smart "closed-loop" system that combines hands-on lab work at DTU Electro with advanced AI from DTU Compute.

“By letting the AI learn from every experiment and suggest the best next steps, we can quickly discover how to build high-efficiency solar cells for indoor use and beyond,” says Associate Professor Mikkel Schmidt, DTU Compute. Read also: AI could speed up the search for better solar cells

Writing Representation and Probabilistic Trajectory Estimation: Modelling Students' Writing Development.

Associate Professor Michael Riis Andersen, DTU Compute, and Associate Professor Kristine Kabel, The Danish School of Education (DPU) – Aarhus University. DKK 4.5 million.

The WRITE-MODE project will develop much-needed robust knowledge about students’ writing development in Grades 4-6, under-researched due to challenges in tracking and quantifying progress over time.

The researchers will integrate expert educational knowledge and advanced data science to capture how linguistic features – ranging from grammar to coherence and voice – emerge and evolve in student texts.

As part of this, they will develop new probabilistic and uncertainty-aware models and extend the statistical Rasch framework to account for uncertainty, temporal dynamics, and writing context, pushing beyond limitations of current methods.

The project unites school writing research, educational linguistics, probabilistic modelling, and uncertainty quantification, and hereby we aim to create a strong interdisciplinary synergy capable of providing new evidence-based insights and advancing student writing research and robust and reliable machine learning," says Associate Professor Michael Riis Andersen, DTU Compute.

STEM NEXT: Spatio Temporal Extreme Rainfall Modelling for Next Generation Hazard Assessment.

Associate Professor Anders Stockmarr, DTU Compute, Technical University of Denmark and Professor Peter Steen Mikkelsen, DTU Sustain, Technical University of Denmark. DKK 4.4 million.

Preparing society for climate change requires much better tools for understanding extreme rainfall than today’s methods, which cannot fully use the rapidly growing mix of historical gauges, modern weather stations, and private sensor networks.

The researchers aim here to establish a new interdisciplinary research direction that positions Denmark at the forefront of next‑generation climate services. By merging advanced statistical modelling with hydrology and atmospheric science, they aim to create a scalable framework that can inform both national and international planning.

They will:

(1) integrate heterogeneous data in a consistent way

(2) capture changing storm behavior

(3) connect large‑scale weather patterns to local extremes

(4) distinguish physical processes from hidden spatial effects

(5) extend the framework across Northern Europe.

The complementary strengths of the two PIs provide a strong foundation for building a long‑term research platform and strengthening future climate services.

“We hope that, within three years, we will have models that significantly improve our ability to support future climate adaptation decisions. We also hope to demonstrate that the methodology can be applied beyond Denmark,” says Anders Stockmarr.

Read also the news: Grant award to improve predictions of extreme rainfall

A Fleet‑Scale Data-driven Framework for Global Wave Sensing (Fleet-GLOWS).

Associate Professor Allan Engsig-Karup, DTU Compute and Associate Professor Ulrik Dam Nielsen, DTU Construct. DKK 4.4 million.

Accurate ocean-wave data are essential for weather forecasting, climate research, and safe, fuel-efficient shipping, yet conventional wave buoys provide sparse coverage. In principal, this project transforms the global merchant fleet into a distributed wave-observing system using ship motion measurements (MRU data) already collected onboard.

Moving beyond single-vessel studies, we analyse MRU data from about 150 container ships in worldwide service and treat wave estimation as a fleet-scale inference problem.

By combining hydrodynamic transfer functions with physics-informed neural networks, transfer learning, and uncertainty quantification, the framework embeds physical constraints while leveraging large-scale data. Aggregating diverse vessel responses reduces bias and uncertainty without new hardware.

The project establishes scalable, self-validating wave estimation with impact on maritime safety, operational efficiency, and improved ocean observations for weather and climate models.

“With large volumes of ship data and physics-informed AI, the aim is to turn an entire fleet of ships into a single network of wave sensors, improving fuel efficiency and safety at sea,” Associate Professor Allan Engsig-Karup.