Спецсеминар Bayes Group: Generative flow networks
02 июня состоится последний спецсеминар учебного года, начало в 18 часов.
Выступит: Александра Волохова (PhD-студентка, Mila — Quebec AI institute, Université de Montréal)
Тема: Generative flow networks
Абстракт: Generative flow networks (GFlowNets) are amortized variational inference algorithms that are trained to sample from unnormalized reward distributions over compositional objects. The key feature of these algorithms is their ability to generate a diverse set of high-reward objects, which is very useful for scientific discovery applications. Also, GFlowNets can accommodate both discrete and continuous properties of the objects. In this talk, I’ll focus on explaining the algorithm and its mathematical foundations, its relation to diffusion models, and show you some examples of its applications.