Bayesian Learning via Neural Schrödinger-Föllmer Flows
This work addresses the problem of efficient Bayesian inference for large-scale data, presenting an incremental improvement over existing methods.
The paper tackles approximate Bayesian inference in large datasets by proposing a stochastic control framework as a finite-time, low-variance alternative to steady-state methods like SGLD, and it adapts theoretical guarantees while connecting to existing VI routines in SDE-based models.
In this work we explore a new framework for approximate Bayesian inference in large datasets based on stochastic control (i.e. Schrödinger bridges). We advocate stochastic control as a finite time and low variance alternative to popular steady-state methods such as stochastic gradient Langevin dynamics (SGLD). Furthermore, we discuss and adapt the existing theoretical guarantees of this framework and establish connections to already existing VI routines in SDE-based models.