Amirhossein Ghaffari

I am a doctoral researcher in Computer Science and Engineering at the Future Computing Group (FCG) within the Centre for Applied Computing, Faculty of Information Technology and Electrical Engineering (ITEE), University of Oulu, Finland. My work lies at the intersection of urban computing, machine learning, and graph neural networks, with a focus on developing Graph Neural Networks-based methods for city-scale predictions and intelligent decision support. Combining ideas from machine learning, network science, and urban analytics, I aim to model dynamic, data-driven systems that enhance urban well-being and sustainability.

From September to November 2026, I am a visiting researcher with the Machine Learning for Smart Mobility (MLSM) group at the Technical University of Denmark (DTU) in Kongens Lyngby, Denmark, hosted by Prof. Francisco Camara Pereira.

Alongside my doctoral work, I have contributed to projects spanning federated learning, smart city optimization, generative AI, agentic AI and machine learning for environmental and communication systems. My background includes computer vision, signal analysis, and predictive modeling using both deep learning and classical ML approaches.

I am passionate about turning heterogeneous data, from sensors, networks, and open urban datasets, into actionable insights through scalable, interpretable, and privacy-preserving AI models.

Before starting my PhD, I worked for about three years as a computer vision and deep learning researcher at Fard Iran Co., designing algorithms for automatic number plate recognition, weigh-in-motion (WIM) systems, and intelligent surveillance, and helping deploy these systems at scale (deployed at more than 70 stations). I hold an M.Sc. from Sharif University of Technology and a B.Sc. from Amirkabir University of Technology.

My current work spans spatio-temporal graph neural networks, traffic and mobility modeling, and broader questions around how heterogeneous urban data can be processed and used efficiently. I am always happy to discuss research, collaboration, or interesting opportunities; feel free to reach out via email or LinkedIn.

news

Sep 08, 2026	Starting a research visit at the Machine Learning for Smart Mobility (MLSM) group, Technical University of Denmark (DTU), Kongens Lyngby, Denmark, hosted by Prof. Francisco Camara Pereira, from September to November 2026.
Aug 20, 2026	Will present our paper When We Talk Morse: Lightweight Parameter Encoding for Secure Federated Learning at the The International Conference on Availability, Reliability and Security (ARES 2026), Linköping, Sweden, on 24 – 27 August 2026.
Apr 30, 2026	Will present our paper Graph-Conditioned Mixture of Graph Neural Network Experts for Traffic Forecasting at the 27th IEEE International Conference on Mobile Data Management (MDM 2026), Athens, Greece, on 29 June – 2 July 2026.

See all 16 news →

selected publications

RedditPersona: A Modular Framework for Community-Conditioned LLM Adaptation from Reddit

Amirhossein Ghaffari, Ali Goodarzi, Huong Nguyen, and 3 more authors

In Proceedings of the 35th ACM International Conference on Information and Knowledge Management (CIKM), 2026

Bib HTML Code

@inproceedings{ghaffari2026redditpersona,
  title = {RedditPersona: A Modular Framework for Community-Conditioned LLM Adaptation from Reddit},
  author = {Ghaffari, Amirhossein and Goodarzi, Ali and Nguyen, Huong and Hosio, Simo and Lovén, Lauri and Gilman, Ekaterina},
  booktitle = {Proceedings of the 35th ACM International Conference on Information and Knowledge Management (CIKM)},
  year = {2026},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  numpages = {5},
  url = {https://doi.org/10.1145/3799682.3840182},
}

Graph-Conditioned Mixture of Graph Neural Network Experts for Traffic Forecasting

Amirhossein Ghaffari, Saeid Sheikhi, and Ekaterina Gilman

In 27th IEEE International Conference on Mobile Data Management (MDM), 2026

Bib HTML Code

@inproceedings{ghaffari2026graphconditioned,
  author = {Ghaffari, Amirhossein and Sheikhi, Saeid and Gilman, Ekaterina},
  title = {Graph-Conditioned Mixture of Graph Neural Network Experts for Traffic Forecasting},
  booktitle = {27th IEEE International Conference on Mobile Data Management (MDM)},
  year = {2026},
}

STRAM: Spatio-Temporal Road-Aware Mapping for Graph Neural Network Prediction

Amirhossein Ghaffari, Huong Nguyen, Lauri Lovén, and 1 more author

Neurocomputing, 2026

Bib HTML Code

@article{ghaffari2026stram,
  author = {Ghaffari, Amirhossein and Nguyen, Huong and Lov{\'e}n, Lauri and Gilman, Ekaterina},
  title = {STRAM: Spatio-Temporal Road-Aware Mapping for Graph Neural Network Prediction},
  journal = {Neurocomputing},
  volume = {696},
  pages = {134161},
  year = {2026},
  issn = {0925-2312},
}

STM-Graph: A Python Framework for Spatio-Temporal Mapping and Graph Neural Network Predictions

Amirhossein Ghaffari, Huong Nguyen, Lauri Lovén, and 1 more author

In Proceedings of the 34th ACM International Conference on Information and Knowledge Management (CIKM), 2025

Bib HTML Code

@inproceedings{ghaffari2025stm,
  title = {STM-Graph: A Python Framework for Spatio-Temporal Mapping and Graph Neural Network Predictions},
  author = {Ghaffari, Amirhossein and Nguyen, Huong and Lov{\'e}n, Lauri and Gilman, Ekaterina},
  booktitle = {Proceedings of the 34th ACM International Conference on Information and Knowledge Management (CIKM)},
  pages = {6377--6381},
  year = {2025},
}

See all 14 publications →