langevin dynamics

This is a technique used in machine learning for sampling from distributions using stochastic gradients, which provides a way to explore the parameter space while incorporating noise, often aimed at achieving a balanced trade-off between exploration and exploitation.

10 papers