My research revolves around interdisciplary applications using structured deep learning. The main focus of my PhD is attempting to speed up space weather simulation that follows the hybrid-Vlasov formalism.
Experience
- Visiting researcher at CMCC building an ML based alternative to the operational numerical Mediterranean forecasting system.
- Visiting researcher at SARAO developing computer vision techniques to deconvolve radio interferometric images.
- Technical student at CERN working with Kubeflow, a cloud native service supporting scalable end-to-end machine learning pipelines.
- Further information in my CV.
Publications
 |
Accurate Mediterranean Sea forecasting via graph-based deep learning Daniel Holmberg, Emanuela Clementi, Italo Epicoco, Teemu Roos preprint, 2025 arxiv | slides |
 |
Global fields of daily accumulation-mode particle number concentrations using in situ observations, reanalysis data and machine learning Aino Ovaska, Elio Rauth, Daniel Holmberg, et al. preprint, 2025 paper |
 |
Learning Developmental Age from 3D Infant Kinetics Using Adaptive Graph Neural Networks Daniel Holmberg, Manu Airaksinen, Viviana Marchi, Andrea Guzzetta, Anna Tuiskula, Leena Haataja, Sampsa Vanhatalo, Teemu Roos IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2025 paper | arxiv |
 |
Regional Ocean Forecasting with Hierarchical Graph Neural Networks Daniel Holmberg, Emanuela Clementi, Teemu Roos In NeurIPS 2024 Workshop on Tackling Climate Change with Machine Learning, 2024 paper | arxiv | slides | poster |
 |
Jet Energy Calibration with Deep Learning as a Kubeflow Pipeline Daniel Holmberg, Dejan Golubovic, Henning Kirschenmann Computing and Software for Big Science, Vol 7, 2023 paper | arxiv | kubecon video | meetup video |
 |
Jet Energy Corrections with Graph Neural Network Regression Daniel Holmberg Master’s Thesis, University of Helsinki, 2022 thesis | slides | poster |
 |
Interatomic Fe–Cr potential for modeling kinetics on Fe surfaces Pekko Kuopanportti, Matti Ropo, Daniel Holmberg, et al. Computational Materials Science, Vol 203, 2022 paper | arxiv |
 |
Potentialmodeller vid simulering av Fe-Cr Daniel Holmberg Bachelor’s Thesis, University of Helsinki, May 2019 thesis | slides |
Talks
| Accurate Mediterranean Sea Forecasting via Graph-based Deep Learning | The Way Forward Workshop: AI/ML in Earth System Science |  |
| Regional Ocean Forecasting with Hierarchical Graph Neural Networks | CMCC Seminar | |
| CERN ML Platform | CMG Group Meeting |  |
| Jet Energy Corrections with GNN Regression using Kubeflow at CERN | KubeCon Valencia |  |
| Jet Energy Corrections with Graph Neural Network Regression | Learning to Discover |  |
| Centralized Management of Your Machine Learning Lifecycle | CERN IT Technical Forum | |
| Jet Energy Corrections with DNN Regression | CMS Machine Learning Forum | |
| Emerging Computing Architectures | Distributed ML Seminar |  |
| Top Quark Mass Measurement | LPC CMS Data Analysis School | ㅤ |