Uppsala, Sweden – On 21 March, Mihai Croicu successfully defended his PhD dissertation, “Forecasting Battles: New Machine Learning Methods for Predicting Armed Conflict.” Croicu’s work, developed within VIEWS, represents a major step forward in ML/AI-powered conflict forecasting.

In the words of Croicu’s opponent, Dr. Thomas Chadefaux, the dissertation:

➡️ Improves forecasting accuracy by integrating real-time event data.
➡️ Develops new methods for handling geospatial uncertainty in conflict datasets.
➡️ Introduces active learning to extract predictive features from text data.
➡️ Shifts focus from locations to actors. Improves escalation predictions.
➡️ Explores media biases in conflict event reporting.

Croicu’s dissertation summary (kappa) provides a compelling 60-page overview of his impressive contributions to the field. 

We extend our gratitude to Thomas Chadefaux from Trinity College Dublin for serving as Croicu´s opponent, as well as Karina Shyrokykh, Sara Stymne and Magnus Öberg, who served on the grading committee.

Last but not least, we thank Croicu’s supervisors Håvard Hegre, Joakim Kreutz, and Nils B. Weidmann for their invaluable guidance throughout his PhD journey.

Huge congratulations again, Mihai – well deserved! 👏