UNIVERSITY OF HERTFORDSHIRE
             COMPUTER SCIENCE RESEARCH COLLOQUIUM

                            presents

     
	"scikit-agent: a Python toolkit for dynamic structural causal games"
	

                        Dr. Sebastian Benthall
                     (International Computer Science Institute, NYU)

                          5 August 2025 (TUESDAY)
                         13:00 -14:00 
                          Room E350

                  Everyone is Welcome to Attend
		          Refreshments will be provided
				  
				                               



Abstract:

Causal Bayesian Networks (CBN) and, more specifically, structural causal models (SCM) are powerful frameworks for representing systems. Structural Causal Games (SCG) (also known as Multi-Agent Influence Diagrams (MAID)) are decision-theoretic, multi-agent variations of SCMs that have recently received renewed interest in artificial intelligence research. Because these models can carefully represent information flows between agents, they are apt for representing sociotechnical systems involving data flows. While SCGs have attractive analytic properties that allow the problem of solving for strategic equilibrium to be efficiently decomposed, nontrivial models can have no closed form analytic solution. We introduce scikit-agent, a new Python toolkit for representing and solving SCGs computationally. This toolkit can solve both static and dynamic SCGs by adopting recent deep learning methods from computational economics. This talk relies on examples from research about privacy engineering and financial regulation.

---------------------------------------------------
Hertfordshire Computer Science Research Colloquium
http://cs-colloq.cs.herts.ac.uk