Silke Kaiser Wins Best Paper Award at ECML PKDD…

The Intelligent Transportation Systems (ITS) group at DTU is pleased to share that Silke Kaiser, Doctoral Researcher in Machine Learning and Sustainable Transport, has received the Best Paper Award at the SoGood Workshop during the ECML PKDD 2025 conference in Porto, Portugal.

Her paper, “Spatio-Temporal Graph Neural Network for Urban Spaces: Interpolating Citywide Traffic Volume”, presents a novel approach to modeling and interpolating citywide traffic volume using spatio-temporal graph neural networks. This research contributes to advancing data-driven methods for understanding and managing urban mobility.

This work was partially conducted during Silke Kaiser’s research visit to the ITS group at DTU, reflecting the value of international collaboration and exchange of ideas in advancing transportation research.

The paper is available on arXiv:
🔗 https://lnkd.in/eHVA3RP5

We extend our congratulations to Silke Kaiser and her co-authors Filipe Rodrigues, Carlos Lima de Azevedo, and Lynn Kaack for this significant recognition.

New Ph.D. graduation in our group: congrats Atefeh!

On November 21st, 2024, Atefeh Hemmati Golsefidi successfully defended her Ph.D. thesis entitled “Adaptive and Flexible Charging Network Expansion with Reinforcement Learning, Optimization and Simulation“.

Congratulations, Atefeh!

She was supervised by Professor Francisco Camara Pereira and Associate Professor  Samitha Samaranayake (Cornell University).

Examiners were Professor Bissan Ghaddar (DTU Management), Associate Professor Andrea Araldo (Institut Polytechnique de Paris – Telecom SudParis) and Assistant Professor Mauro Salazar (Eindhoven University of Technology).

The defense session was chaired by Associate Professor Filipe Rodrigues.

For more information, feel free to contact Atefeh via her Linkedin

Seminar by Ricardo Daziano

We had the pleasure of hosting Ricardo Daziano, Associate Professor at Cornell University, for an insightful seminar on Neural Networks for Choice Analysis: Enhancing Behavioral Regularity and Interpretability.

Professor Daziano discussed innovative approaches to integrating machine learning with econometric principles, focusing on improving behavioral consistency in travel demand forecasting and incorporating network effects into discrete choice models using Graph Neural Networks (GNNs).

Below you can find the abstract and more information about Ricardo’s research interests.

Abstract: This talk explores the integration of machine learning with econometric principles, focusing on behavioral regularization in travel demand modeling. Two case studies will illustrate the need for combining behavioral regularization with advanced machine learning techniques to ensure that models are both theoretically sound and practically useful in travel decision-making contexts. The first case study examines monotonicity in travel demand forecasts from deep neural networks (DNNs). Despite the predictive strength of DNNs, predictions can fail to maintain behavioral consistency. New metrics for assessing monotonicity will be discussed, alongside a constrained optimization framework that exploits gradient regularizers. Applied to travel survey data from Chicago and London, the proposed approach improves DNN behavioral regularity while preserving predictive accuracy, with notable enhancements in smaller samples and out-of-domain scenarios. The second case study introduces a novel graph neural network (GNN) architecture designed to incorporate network effects into discrete choice models, while keeping parametric interpretability. Traditional discrete choice models typically overlook or neglect peer influence, but the proposed GNN approach effectively integrates peer effects, resulting in superior predictive performance and greater interpretability. When applied to New York City commuting data, the GNN model provides valuable economic insights, such as the value of time (VOT), demonstrating its effectiveness in capturing network effects and offering robust policy implications.

Bio: Ricardo A. Daziano, PhD in Economics and Associate Professor of Civil and Environmental Engineering at Cornell University, is a choice modeler working on deriving Bayesian and semiparametric estimators for microeconometric models of customer decisions in engineering contexts. Daziano’s research has a unique focus on applied econometrics of consumer choices around energy efficiency. His technical goal is to derive and apply preference estimators that are econometrically, behaviorally, and computationally superior to extant tools. His empirical goal is to better understand the interaction of consumer behavior and engineering, investment, and policy choices for energy-efficient technologies and sustainability, especially in transportation. In his transdisciplinary career, Daziano has been particularly interested in modeling the socio-technical transition to electric vehicles (EVs), and has used non-market valuation techniques to ascertain preferences for EV features, as well as preferences for EV charging and electricity contracts.

New members in our group: Welcome Francisco, João and…

Francisco Madaleno, João Böger, and Oskar Bohn Lassen have joined our group as PhD students, working on the ‘APEX – Artificial Intelligence for Policy Excellence in the Climate Crisis’ project, under the supervision of Professor Francisco Camara Pereira, along with three colleagues: Associate Professor Carlos Lima Azevedo, Assistant Professor Guido Cantelmo (Transportation Systems Modelling – DTU), and Associate Professor Filipe Rodrigues. Their work in APEX focuses on developing various machine learning-based approximations to simulation tools, which aim to significantly accelerate their runtimes by several orders of magnitude. APEX will be applied in crucial domains such as transport, energy, and the environment, areas that have significant societal impact. Committed to the principles of human-aligned AI, APEX ensures that its artificial intelligence operates with a focus on transparency, fairness, and accountability, aligning technological advancements with human values and ethical considerations.

Francisco holds a Master’s degree in Aerospace Engineering, specializing in Aerodynamics. After completing his studies, he worked as an Applied Scientist at Zalando, focusing on price optimization and item flow distribution. João holds a Bachelor’s degree from UFSC and a Master’s degree from IFT-Unesp in Physics, where he specialized in Theoretical, Particle, and Computational Physics. Following his studies, he worked on Computer Vision and Deep Learning applications in Geophysical Sciences at LTrace – Geophysical Solutions. Oskar holds a Master’s degree in Business Analytics from DTU, where he focused on machine learning, operations research, and differential geometry. He brings two years of experience as a management consultant at McKinsey & Company, where he worked on strategy and digital topics. He is excited to apply his industry experience and academic expertise to the fields of AI and machine learning.

New member in our group: welcome Julia!

Julia Guggenberger has joined our group as a PhD student, working on the ‘Green Token Project: Digital Tradable Credits for Sustainable Personal Mobility’ under the supervison of Associate Professor Carlos Lima Azevedo and Assistant Professor Ravi Seshadri (Transportation Systems Modelling group – DTU) and Doctor Biagio Ciuffo (European Commission Joint Research Centre). This project is a collaboration between DTU and the JRC and focuses on understanding how individuals behave when traveling under a credit-based scheme for sustainable mobility.

Julia holds a Master’s degree in Applied Economics from the University of Innsbruck, where she specialized in Behavioral Economics, Econometrics, and Statistics. Following her studies, she completed a traineeship at the European Central Bank and has now chosen to pursue an academic career, with a particular interest in environmental topics.

We wish her a fun and productive time during her stay with us! 😊

New research grant: congrats Rico!

Recently, Assistant Professor Rico Krueger from our group received the prestigious European Research Council Starting Grant for his project, IMMERSION: Explaining Human Decision-Making by Combining Choice and Process Data!

Over the five-year duration of the IMMERSION project, Rico and his team will establish a novel framework to understand and predict human behavior in hypothetical and futuristic scenarios. This includes developing new models that integrate choice and process data, designing new statistical inference procedures tailored to these models, and creating innovative methods for collecting rich behavioral data through immersive experiments using extended reality.

The project will also involve extensive empirical work, with significant implications for human-centered design in future transport and infrastructure systems. The European Research Council (ERC) recently announced the award of 494 Starting Grants, distributing nearly €780 million to young scientists and scholars across Europe.

This ERC funding is part of the EU Horizon Europe programme, supporting cutting-edge research across diverse fields. The ERC is Europe’s premier funding body for frontier research, offering grants to researchers of all nationalities and ages across the continent.

Out of 3,474 proposals evaluated by peer-review panels composed of internationally renowned researchers, 14.2% were selected for funding, highlighting the significance and quality of these projects.

Huge congratulations to Rico!

New Ph.D. graduation in our group: congrats Frederik!

On August 23rd, 2024, Frederik Boe Hüttel successfully defended his Ph.D. thesis entitled “Deep Bayesian Modelling for Uncertainty Estimation in Transportation Systems“.

Congratulations, Frederik!

He was supervised by Professor Francisco Camara Pereira and Associate Professor  Filipe Rodrigues.

Examiners were Associate Professor Michael Riis Andersen (DTU Compute), Professor Constantinos Antoniou (Technical University of Munich) and Professor Latifa Oukhellou (Gustave Eiffel University).

The defense session was chaired by Associate Professor Jesper Bláfoss Ingvardson.

List of Papers:

• Paper 1: Hüttel, F. B., Peled, I., Rodrigues, F., & Pereira, F. C. (2022). Modeling censored mobility demand through censored quantile regression neural networks. IEEE Transactions on Intelligent Transportation Systems, 23(11), 21753-21765.

• Paper 2: Hüttel, F. B., Rodrigues, F., & Pereira, F. C. (2023). Mind the gap: Modelling difference between censored and uncensored electric vehicle charging demand. Transportation Research Part C: Emerging Technologies, 153, 104189.

• Paper 3: Hüttel, F. B., Rodrigues, F., & Pereira, F. C. (2023). Deep Evidential Learning for Bayesian Quantile Regression. arXiv preprint arXiv:2308.10650.

• Paper 4: Hüttel, F. B., Riis, C., Rodrigues, F., & Pereira, F. C. (2024). Bayesian Active Learning for Censored Regression. arXiv preprint arXiv:2402.11973.

For more information, feel free to contact Frederik via his Linkedin page.

New visitor in our group: welcome Shadi!

Recently, Shadi Haj Yahia began a three-month visit to our group to collaborate with Prof. Carlos Lima Azevedo on leveraging advanced modeling techniques. This visit is part of the EIT Urban Mobility Doctoral Training Network program, which aligns perfectly with his ambition to bridge the gap between theoretical and practical research in transport planning.

Shadi is a Ph.D. candidate in Transportation Engineering from the Technion – Israel Institute of Technology. His research focuses on expanding machine learning approaches for travel behavior modeling. Among his works are the integration of machine learning methods in travel demand models and enhancing interpretability by incorporating domain knowledge into deep neural networks for better policy-making decisions. Additionally, he developed a grammatical evolution approach to data-driven utility specification for discrete choice models, combining machine learning with domain knowledge to automate utility specification, ensuring interpretability and predictive performance.

We hope he has a great time staying with us!

New member in our group: welcome Miguel!

Miguel Costa recently joined our group as a Postdoc Researcher and will be working on the “MA’AT – Maximising well-being with AI under deep climate turmoil” project under the supervision of Francisco Câmara Pereira. His research at DTU will focus on the combination of transport simulation and reinforcement learning to uncover optimal policies to maximize individuals’ wellbeing under flooding events.

Miguel recently finished his PhD in Transportation Systems from Instituto Superior Técnico (IST) with Distinction and Honour, which focused on using different machine learning tools to study objective and subjective cycling safety, and the relation between the two. Miguel’s research interests are intelligent transportation systems, machine learning applied to transportation scenarios, and cycling safety.

We wish him a fun and productive time during his stay with us.

New visitor in our group: welcome Silke!

We’re delighted to welcome Silke Kaiser for a three-month visit to work with Prof. Lima Azevedo. Coming from the Hertie School in Berlin, Silke is well-versed in sustainable transport and the intelligent use of machine learning.

With a master’s degree from Sorbonne-Panthéon and SciencesPo Paris, she’s tackling an interesting challenge: estimating the volume of bicycles in cities using different data sources and machine learning algorithms. This effort is key to planning more bicycle-friendly urban environments and she wants to contribute to improving the quality of life in urban spaces.

Silke’s stay promises great discussions and insights on how to make cities greener and more efficient. Welcome Silke!