👋 Hi there!

I’m Giovanni Briglia. Since November 2024, I’ve been a PhD student in the National PhD in AI, supervised by Franco Zambonelli and Stefano Mariani.

🎯 Research Overview

My work focuses on Causality-Driven Reinforcement Learning across both single- and multi-agent domains, investigating how causal reasoning can be efficiently represented and integrated into RL to enhance generalization, sample efficiency, and the interpretability of decision-making processes.

🔍 Research Questions

• Representation — What form of representation (possibly learned and continuously updated in real time) best captures and infers transition dynamics and mechanisms both effectively and explainably? How can such representations be reused across environments, tasks, and policies?

• Integration — To what extent, and through which mechanisms, can causal knowledge be integrated into RL so that it becomes an active component of policy optimization and value estimation? At which stage(s) of the RL process is this integration most effective?

• Multi-Agent — How can causality-driven RL methods be extended to multi-agent settings, where cooperation and competition introduce additional layers of complexity and interdependence?

🧠 Main Interests

Reinforcement Learning · Causal Inference · Probabilistic Knowledge Representation through Bayesian and Causal (World) Models

📄 CV

For more details, check my CV.