Articles by M. V. Tretyakov
Neural variance reduction for stochastic differential equations
This paper proposes the use of neural stochastic differential equations as a means to learn approximately optimal control variates, reducing variance as trajectories of the SDEs are simulated.