Dynamic Message Passing and its Applications to Spreading Processes on Networks

Prelegent(ci)

Mateusz Wiliński

Afiliacja

Instytucie Badawczym IDEAS.

Język referatu

angielski

Termin

4 marca 2026 14:15

Pokój

p. 5070

Seminarium

Seminarium Zakładu Biomatematyki i Teorii Gier

---

Spreading models are routinely used to generate predictions for a plethora of diffusion processes on networks, whereby infectious diseases, opinions, or failures propagate in natural, social, and technological systems. In my talk I will describe the dynamic message passing, an inference method derived from belief propagation (cavity method). I will focus on its applications, especially in the regime of uncertainty and/or partial observation. I will also discuss its limitations connected to the loopinnes of the graph, as well as the unidirectional nature of the process for which it is derived. Finally, I will propose some interesting future directions, which could help to overcome the limitations.

---