Events

Diffusion processes have become one of the most significant mathematical frameworks in machine learning (ML), exerting considerable influence across various subfields, including generative modelling, sampling techniques, and nonconvex optimization. The exceptional empirical success of diffusion-based approaches has prompted a growing interest in exploring their mathematical foundations and has driven the development of novel diffusion-based algorithms. In the context of finance, one could argue that applications of diffusion-based generative models can benefit tasks such as scenario generation for risk management purposes and generation of approximate solutions to complex pricing models, such as those used for path-dependent options or exotic derivatives. We will review some recent progress in diffusion-based stochastic optimizers and diffusion-based/score-based generative models.