Blog

Climate AI Nordics. I have been featured! Proof that if you stare at enough flood maps for long enough, people will eventually write about you. Or convincing neural networks to identify floods and training RL agents to pretend they understand transport and water dynamics may be even better than staring at flood maps. I will take what I can get.

BOKU University (Vienna, Austria). Following a short visit in Vienna, I gave a seminar about the MAAT project and how we can use reinforcement learning to uncover optimal policy pathways for increasing transportation systems resilience to climate change.

Active Mobility Workshop: "Are Cities Ready to Bike? Cities for Cycling & Urban Cycling Readiness". I presented some of my past research on what is currently known about the relation between objective and subjective cycling safety and how machine learning can play a role in uncovering the relation between the two types of safety at Masaryk University in Brno.

EIT Urban Mobility Community Days 2026. I presented the latest results of the MA'AT project, showing how we can use reinforcement learning to uncover optimal policy pathways for increasing transportation systems to climate change.

AI for Climate and Conservation Workshop at EurIPS 2025. Arthur Vandervoort presented some preliminary results of our MA'AT framework to showcase climate adaptation strategies targeting quality of life impacts.

The 2025 Conference on Empirical Methods in Natural Language Processing. Xixi Wang and I presented our graph-based framework to better extract information from multi-table databases using LLMs that leverage human curated relational knowledge.

AI in Science 2025. Arthur Vandervoor presented our MA'AT framework.

10th International Conference on Intelligent Transportation Engineering (ICITE) 2025. I gave a keynote talk on how machine learning can be used to model and map objective and subjective cycling safety.

Danish Meteorological Institute's 2025 Climate Symposium. Arthur Vandervoor and Martin Drews presented our work's preliminary results about the MA'AT framework.

Université Gustave Eiffel (Paris, France). I gave a seminar talk on how machine learning can be used in climate adaptation planning when considering economic impacts and quality of life impacts.

EIT Urban Mobility DTN Annual Forum 2025. I gave a keynote talk on how machine learning can be used to aid in climate adaptation planning.

Transport Reviews. My article, coauthored with Manuel Marques, Felix W. Siebert, Carlos L. Azevedo, and Filipe Moura is out! Here, I do a scoping review of the literature on the relation between objective cycling safety and subjective cycling safety, drawing attention to the fact that the relation between the two is very complex and studies have found contradicting results on the possible relation between the two. We need more research to better characterize and understand the relation between the two types of safety.

Cities. My article, coauthored with Gabriel Valença, Cristian Adorean, and Filipe Moura, is finally out! We analyze the relation between circuity and mobility cultures in 41 european cities, showing that circuity varies across city types and their mobility cultures.

Tackling Climate Change with Machine Learning workshop at ICLR 2025. Myself and Arthur Vandervoort presented the overall concept behind the MA'AT framework. We won the Best Proposal Award!

IEEE Transactions on Intelligent Transportation Systems. My article, coauthored with Manuel Marques, Carlos L. Azevedo, Felix W. Siebert, and Filipe Moura is out! We designed a Siamese convolutional neural network capable of taking pairwise comparisons of images to estimate perception of cycling safety scores for urban environments.