Portrait of Giovanni Briglia
Giovanni Briglia
PhD student, National PhD in AI · UNIMORE · visiting the University of Oxford

Agents that learn what their actions cause.

I work on causality-driven reinforcement learning for single- and multi-agent systems. I study how agents can learn causal models while they act, and use them to explore faster, transfer to new tasks and explain their decisions.

Supervised by Franco Zambonelli and Stefano Mariani at UNIMORE. At Oxford I’m hosted by Alessandro Abate and Francesco Fabiano.

Time-unrolled causal graph of two agents. The state at time t points to the actions of agent 1 and agent 2 and to the next state, and both actions point to the next state. An intervention on agent 1's action cuts the edge from the state to that action. TIME t TIME t+1 st a¹ a² st+1 do(a¹) state agent 2 next state agent 1, intervened agent 2 environment
Fig. 1 · Intervening on agent 1, do(a¹), cuts its dependence on the current state. A causal model predicts how that choice changes the next joint state.
Award
Best Paper Award, AREA Workshop @ ECAI 2025
Oral
Full paper with oral presentation at AAMAS 2026
Doctoral Consortium
AAMAS 2026, mentored by Christopher Amato
Visiting
University of Oxford, from September 2026
Research agenda

Three questions shape my PhD

They go from how causal knowledge is represented, to where it enters the learning loop, to what changes when many agents share one environment.

Publications

Papers and preprints

Google Scholar profile →
2026
Diagram of the GNN-derived heuristic pipeline

Scaling Multi-Agent Epistemic Planning through GNN-Derived Heuristics

G. Briglia, F. Fabiano, S. Mariani
AAMAS 2026 · Full paper
Oral arXiv Slides
BibTeX
@inproceedings{brigliascaling,
  title={Scaling Multi-Agent Epistemic Planning through GNN-Derived Heuristics},
  author={Briglia, Giovanni and Fabiano, Francesco and Mariani, Stefano},
  booktitle={The 25th International Conference on Autonomous Agents and Multi-Agent Systems},
  year={2026}
}
2026
Doctoral Consortium poster

Causal Learning and Reasoning in Multi-Agent Reinforcement Learning

G. Briglia
AAMAS 2026 · Doctoral Consortium
ACM DL Poster
BibTeX
@inproceedings{briglia2026causal,
  title={Causal Learning and Reasoning in Multi-Agent Reinforcement Learning},
  author={Briglia, Giovanni},
  booktitle={Proc. of the 25th International Conference on Autonomous Agents and Multiagent Systems},
  pages={3975--3977},
  year={2026}
}
2026
Results of causal value estimation

Causal Models Improve Reinforcement Learning for Pervasive and Robotic Tasks

G. Briglia, S. Mariani, F. Zambonelli
CoMoRe-AI Workshop @ IEEE PerCom 2026 · in press
Slides
BibTeX
@inproceedings{briglia2026pervasive,
  title={Causal Models Improve Reinforcement Learning for Pervasive and Robotic Tasks},
  author={Briglia, Giovanni and Mariani, Stefano and Zambonelli, Franco},
  booktitle={2026 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops)},
  year={2026},
  organization={IEEE Computer Society}
}
2026
Taxonomy of causal learning in multi-agent systems

Causal Learning and Reasoning in Multi-Agent Systems: Goals, Issues, and Taxonomy

S. Mariani, G. Briglia, A. Meyer-Vitali, M. Klusch, F. Zambonelli
CLaRAMAS Workshop @ AAMAS 2026 · in press
OpenReview
BibTeX
@inproceedings{mariani2026taxonomy,
  author={Mariani, Stefano and Briglia, Giovanni and Meyer-Vitali, Andr{\'e} and Klusch, Matthias and Zambonelli, Franco},
  title={Causal Learning and Reasoning in Multi-Agent Systems: Goals, Issues, and Taxonomy},
  booktitle={Proceedings of the 1st International Workshop on Causal Learning and Reasoning in Agents and Multiagent Systems, CLaRAMAS},
  publisher={Springer},
  year={2026}
}
2025
Multi-robot flocking simulation

Towards Safe Action Policies in Multi-robot Systems with Causal Reinforcement Learning

G. Briglia, S. Mariani, F. Zambonelli
AREA Workshop @ ECAI 2025
Best Paper Award Springer Slides
BibTeX
@inproceedings{briglia2025towards,
  title={Towards Safe Action Policies in Multi-robot Systems with Causal Reinforcement Learning},
  author={Briglia, Giovanni and Mariani, Stefano and Zambonelli, Franco},
  booktitle={Workshop on Agents and Robots for reliable Engineered Autonomy},
  pages={51--71},
  year={2025},
  organization={Springer}
}
2025
Overview figure of causal multi-agent RL

A Roadmap Towards Improving Multi-Agent Reinforcement Learning With Causal Discovery And Inference

G. Briglia, S. Mariani, F. Zambonelli
arXiv preprint 2503.17803
arXiv
BibTeX
@article{briglia2025roadmap,
  title={A Roadmap Towards Improving Multi-Agent Reinforcement Learning With Causal Discovery And Inference},
  author={Briglia, Giovanni and Mariani, Stefano and Zambonelli, Franco},
  journal={arXiv preprint arXiv:2503.17803},
  year={2025}
}
2024
Agents exploring a grid maze

Improving Reinforcement Learning-Based Autonomous Agents with Causal Models

G. Briglia, M. Lippi, S. Mariani, F. Zambonelli
PRIMA 2024 · Kyoto
Slides Blog post
BibTeX
@inproceedings{briglia2024improving,
  title={Improving Reinforcement Learning-Based Autonomous Agents with Causal Models},
  author={Briglia, Giovanni and Lippi, Marco and Mariani, Stefano and Zambonelli, Franco},
  booktitle={International Conference on Principles and Practice of Multi-Agent Systems},
  pages={267--283},
  year={2024},
  organization={Springer}
}
2024
Bearing fault detection pipeline

Bearing Fault Detection and Recognition From Supply Currents With Decision Trees

G. Briglia, F. Immovilli, M. Cocconcelli, M. Lippi
IEEE Access, vol. 12
DOI
BibTeX
@article{10376052,
  author={Briglia, Giovanni and Immovilli, Fabio and Cocconcelli, Marco and Lippi, Marco},
  journal={IEEE Access},
  title={Bearing Fault Detection and Recognition From Supply Currents With Decision Trees},
  year={2024},
  volume={12},
  pages={12760-12770},
  doi={10.1109/ACCESS.2023.3348245}
}
2023
Cross-load generalization results

Cross-Load Generalization of Bearing Fault Recognition with Decision Trees

G. Briglia, F. Immovilli, M. Cocconcelli, M. Lippi
ICSRS 2023
DOI
BibTeX
@inproceedings{10381353,
  author={Briglia, Giovanni and Immovilli, Fabio and Cocconcelli, Marco and Lippi, Marco},
  booktitle={2023 7th International Conference on System Reliability and Safety (ICSRS)},
  title={Cross-Load Generalization of Bearing Fault Recognition with Decision Trees},
  year={2023},
  pages={400-406},
  doi={10.1109/ICSRS59833.2023.10381353}
}

News

Sep 2026
I was hosted by Prof. Alessandro Abate for an OXCAV seminar, presenting When, Why, and How Causality is needed in Reinforcement Learning!
Sep 2026
I am attending the Erlangen AI Conference hosted at the Mathematical Institute of the University of Oxford! I am presenting my vision of causality in MARL :)
Jun 2026
I presented my PhD Thesis outline “Causal Learning and Reasoning in Multi-Agent Reinforcement Learning” at the AAMAS 2026 Doctoral Consortium. Poster here. It was also a pleasure to have had Christopher Amato as a mentor.
Jun 2026
I presented our work “Scaling Multi-Agent Epistemic Planning through GNN-Derived Heuristics” at AAMAS Conference 2026, in the Planning and Learning track. Slides here.
May 2026
I am excited to announce that, starting on September 1st, I will begin a PhD Visiting Research period at the University of Oxford, where I will be hosted by Alessandro Abate and Francesco Fabiano!!
Jan 2026
Our paper Causal Models Improve Reinforcement Learning for Pervasive and Robotic Tasks has been accepted at the CoMoRe-AI 2026 Workshop, within the IEEE PerCom Conference. Check out the presentation slides here.
Older news (8)
Dec 2025
Our paper “Scaling Multi-Agent Epistemic Planning through GNN-Derived Heuristics” has been accepted as a full paper at AAMAS 2026, with an oral presentation!! 🚀🚀
Oct 2025
🚀 Our paper, “Towards Safe Action Policies in Multi-robot Systems with Causal Reinforcement Learning”, received the Best Paper Award at the AREA Workshop @ ECAI 2025! 🎉
Oct 2025
🚀 Our paper “Towards Safe Action Policies in Multi-robot Systems with Causal Reinforcement Learning” is now published! Read it on Springer. 🎤 I’ll be presenting this work at AREA @ ECAI 2025 — check out the presentation slides here. 🎉
Sep 2025
Excited to share our new preprint: How to Scale Multi-Agent Epistemic Planning with GNNs!! Now available on arXiv !!
Mar 2025
New pre-print on Causal MARL is now available on arxiv !!
Dec 2024
Christmas game early 🎅: I am excited to share that I have been selected to participate in the Data Study Group this January-February, organized by The Alan Turing Institute.
Sep 2024
Our paper, “Improving Reinforcement Learning-based Autonomous Agents with Causal Models,” has been accepted as a regular paper at the 25th International Conference on Principles and Practice of Multi-Agent Systems (PRIMA), which will take place in Kyoto, Japan, from November 18-24, 2024.
Aug 2024
Site online!!

Talks

Sep 2026 · OXCAV seminar, University of Oxford
When, Why, and How Causality Is Needed in Reinforcement Learning
Sep 2026 · Erlangen AI Hub Conference, Oxford
May 2026 · AAMAS 2026 Doctoral Consortium · poster
Oct 2025 · AREA Workshop @ ECAI 2025, Bologna