Solving Schrödinger Bridges via Maximum Likelihood

Francisco Vargas, Pierre Thodoroff, Austen Lamacraft, Neil D. Lawrence
Entropy, 23(9):1134, 2021.

Abstract

The Schrödinger bridge problem (SBP) finds the most likely stochastic evolution between two probability distributions given a prior stochastic evolution. As well as applications in the natural sciences, problems of this kind have important applications in machine learning such as dataset alignment and hypothesis testing. Whilst the theory behind this problem is relatively mature, scalable numerical recipes to estimate the Schrödinger bridge remain an active area of research. Our main contribution is the proof of equivalence between solving the SBP and an autoregressive maximum likelihood estimation objective. This formulation circumvents many of the challenges of density estimation and enables direct application of successful machine learning techniques. We propose a numerical procedure to estimate SBPs using Gaussian process and demonstrate the practical usage of our approach in numerical simulations and experiments.

Cite this Paper

BibTeX
@Article{Vargas-solving21, title = {Solving Schrödinger Bridges via Maximum Likelihood}, author = {Vargas, Francisco and Thodoroff, Pierre and Lamacraft, Austen and Lawrence, Neil D.}, journal = {Entropy}, year = {2021}, volume = {23}, number = {9}, doi = {10.3390/e23091134}, pdf = {https://www.mdpi.com/1099-4300/23/9/1134/pdf}, url = {http://inverseprobability.com/publications/solving-schroedinger-bridges-via-maximum-likelihood.html}, abstract = {The Schrödinger bridge problem (SBP) finds the most likely stochastic evolution between two probability distributions given a prior stochastic evolution. As well as applications in the natural sciences, problems of this kind have important applications in machine learning such as dataset alignment and hypothesis testing. Whilst the theory behind this problem is relatively mature, scalable numerical recipes to estimate the Schrödinger bridge remain an active area of research. Our main contribution is the proof of equivalence between solving the SBP and an autoregressive maximum likelihood estimation objective. This formulation circumvents many of the challenges of density estimation and enables direct application of successful machine learning techniques. We propose a numerical procedure to estimate SBPs using Gaussian process and demonstrate the practical usage of our approach in numerical simulations and experiments. } }
Endnote
%0 Journal Article %T Solving Schrödinger Bridges via Maximum Likelihood %A Francisco Vargas %A Pierre Thodoroff %A Austen Lamacraft %A Neil D. Lawrence %J Entropy %D 2021 %F Vargas-solving21 %R 10.3390/e23091134 %U http://inverseprobability.com/publications/solving-schroedinger-bridges-via-maximum-likelihood.html %V 23 %N 9 %X The Schrödinger bridge problem (SBP) finds the most likely stochastic evolution between two probability distributions given a prior stochastic evolution. As well as applications in the natural sciences, problems of this kind have important applications in machine learning such as dataset alignment and hypothesis testing. Whilst the theory behind this problem is relatively mature, scalable numerical recipes to estimate the Schrödinger bridge remain an active area of research. Our main contribution is the proof of equivalence between solving the SBP and an autoregressive maximum likelihood estimation objective. This formulation circumvents many of the challenges of density estimation and enables direct application of successful machine learning techniques. We propose a numerical procedure to estimate SBPs using Gaussian process and demonstrate the practical usage of our approach in numerical simulations and experiments.
RIS
TY - JOUR TI - Solving Schrödinger Bridges via Maximum Likelihood AU - Francisco Vargas AU - Pierre Thodoroff AU - Austen Lamacraft AU - Neil D. Lawrence DA - 2021/08/31 ID - Vargas-solving21 VL - 23 IS - 9 SP - 1134 DO - 10.3390/e23091134 L1 - https://www.mdpi.com/1099-4300/23/9/1134/pdf UR - http://inverseprobability.com/publications/solving-schroedinger-bridges-via-maximum-likelihood.html AB - The Schrödinger bridge problem (SBP) finds the most likely stochastic evolution between two probability distributions given a prior stochastic evolution. As well as applications in the natural sciences, problems of this kind have important applications in machine learning such as dataset alignment and hypothesis testing. Whilst the theory behind this problem is relatively mature, scalable numerical recipes to estimate the Schrödinger bridge remain an active area of research. Our main contribution is the proof of equivalence between solving the SBP and an autoregressive maximum likelihood estimation objective. This formulation circumvents many of the challenges of density estimation and enables direct application of successful machine learning techniques. We propose a numerical procedure to estimate SBPs using Gaussian process and demonstrate the practical usage of our approach in numerical simulations and experiments. ER -
APA
Vargas, F., Thodoroff, P., Lamacraft, A. & Lawrence, N.D.. (2021). Solving Schrödinger Bridges via Maximum Likelihood. Entropy 23(9):1134 doi:10.3390/e23091134 Available from http://inverseprobability.com/publications/solving-schroedinger-bridges-via-maximum-likelihood.html.

Related Material