Timezone: »
More than twenty years after its introduction, Annealed Importance Sampling (AIS) remains one of the most effective methods for marginal likelihood estimation. It relies on a sequence of distributions interpolating between a tractable initial distribution and the target distribution of interest which we simulate from approximately using a non-homogeneous Markov chain. To obtain an importance sampling estimate of the marginal likelihood, AIS introduces an extended target distribution to reweight the Markov chain proposal. While much effort has been devoted to improving the proposal distribution used by AIS, by changing the intermediate distributions and corresponding Markov kernels, an underappreciated issue is that AIS uses a convenient but suboptimal extended target distribution. This can hinder its performance. We here leverage recent progress in score-based generative modeling (SGM) to approximate the optimal extended target distribution for AIS proposals corresponding to the discretization of Langevin and Hamiltonian dynamics. We demonstrate these novel, differentiable, AIS procedures on a number of synthetic benchmark distributions and variational auto-encoders.
Author Information
Arnaud Doucet (Oxford)
Will Grathwohl (Deepmind)
Alexander Matthews (DeepMind)
Heiko Strathmann (Google Deepmind)
More from the Same Authors
-
2022 : Spectral Diffusion Processes »
Angus Phillips · Thomas Seror · Michael Hutchinson · Valentin De Bortoli · Arnaud Doucet · Emile Mathieu -
2023 Poster: Trans-Dimensional Generative Modeling via Jump Diffusion Models »
Andrew Campbell · William Harvey · Christian Weilbach · Valentin De Bortoli · Thomas Rainforth · Arnaud Doucet -
2023 Poster: Diffusion Schrödinger Bridge Matching »
Yuyang Shi · Valentin De Bortoli · Andrew Campbell · Arnaud Doucet -
2023 Poster: Marginal Density Ratio for Off-Policy Evaluation in Contextual Bandits »
Muhammad Faaiz Taufiq · Arnaud Doucet · Rob Cornish · Jean-Francois Ton -
2023 Poster: Tree-Based Diffusion Schrödinger Bridge with Applications to Wasserstein Barycenters »
Maxence Noble · Valentin De Bortoli · Arnaud Doucet · Alain Durmus -
2023 Poster: A Unified Framework for U-Net Design and Analysis »
Christopher Williams · Fabian Falck · George Deligiannidis · Chris C Holmes · Arnaud Doucet · Saifuddin Syed -
2023 Poster: DISCS: A Benchmark for Discrete Sampling »
Katayoon Goshvadi · Haoran Sun · Xingchao Liu · Azade Nova · Ruqi Zhang · Will Grathwohl · Dale Schuurmans · Hanjun Dai -
2023 Poster: Alpha-divergence Variational Inference Meets Importance Weighted Auto-Encoders: Methodology and Asymptotics »
Kamélia Daudel · Joe Benton · Yuyang Shi · Arnaud Doucet -
2022 Spotlight: Lightning Talks 1A-4 »
Siwei Wang · Jing Liu · Nianqiao Ju · Shiqian Li · Eloïse Berthier · Muhammad Faaiz Taufiq · Arsene Fansi Tchango · Chen Liang · Chulin Xie · Jordan Awan · Jean-Francois Ton · Ziad Kobeissi · Wenguan Wang · Xinwang Liu · Kewen Wu · Rishab Goel · Jiaxu Miao · Suyuan Liu · Julien Martel · Ruobin Gong · Francis Bach · Chi Zhang · Rob Cornish · Sanmi Koyejo · Zhi Wen · Yee Whye Teh · Yi Yang · Jiaqi Jin · Bo Li · Yixin Zhu · Vinayak Rao · Wenxuan Tu · Gaetan Marceau Caron · Arnaud Doucet · Xinzhong Zhu · Joumana Ghosn · En Zhu -
2022 Spotlight: Conformal Off-Policy Prediction in Contextual Bandits »
Muhammad Faaiz Taufiq · Jean-Francois Ton · Rob Cornish · Yee Whye Teh · Arnaud Doucet -
2022 Poster: Conformal Off-Policy Prediction in Contextual Bandits »
Muhammad Faaiz Taufiq · Jean-Francois Ton · Rob Cornish · Yee Whye Teh · Arnaud Doucet -
2022 Poster: A Continuous Time Framework for Discrete Denoising Models »
Andrew Campbell · Joe Benton · Valentin De Bortoli · Thomas Rainforth · George Deligiannidis · Arnaud Doucet -
2022 Poster: A Multi-Resolution Framework for U-Nets with Applications to Hierarchical VAEs »
Fabian Falck · Christopher Williams · Dominic Danks · George Deligiannidis · Christopher Yau · Chris C Holmes · Arnaud Doucet · Matthew Willetts -
2022 Poster: Riemannian Score-Based Generative Modelling »
Valentin De Bortoli · Emile Mathieu · Michael Hutchinson · James Thornton · Yee Whye Teh · Arnaud Doucet -
2022 Poster: Learning to Navigate Wikipedia by Taking Random Walks »
Manzil Zaheer · Kenneth Marino · Will Grathwohl · John Schultz · Wendy Shang · Sheila Babayan · Arun Ahuja · Ishita Dasgupta · Christine Kaeser-Chen · Rob Fergus -
2022 Poster: Towards Learning Universal Hyperparameter Optimizers with Transformers »
Yutian Chen · Xingyou Song · Chansoo Lee · Zi Wang · Richard Zhang · David Dohan · Kazuya Kawakami · Greg Kochanski · Arnaud Doucet · Marc'Aurelio Ranzato · Sagi Perel · Nando de Freitas -
2019 : Invited talk 2 »
Alexander Matthews -
2019 Poster: Augmented Neural ODEs »
Emilien Dupont · Arnaud Doucet · Yee Whye Teh -
2018 Poster: Hamiltonian Variational Auto-Encoder »
Anthony Caterini · Arnaud Doucet · Dino Sejdinovic -
2017 Poster: Filtering Variational Objectives »
Chris Maddison · John Lawson · George Tucker · Nicolas Heess · Mohammad Norouzi · Andriy Mnih · Arnaud Doucet · Yee Teh -
2017 Poster: Clone MCMC: Parallel High-Dimensional Gaussian Gibbs Sampling »
Andrei-Cristian Barbos · Francois Caron · Jean-François Giovannelli · Arnaud Doucet -
2015 Workshop: Scalable Monte Carlo Methods for Bayesian Analysis of Big Data »
Babak Shahbaba · Yee Whye Teh · Max Welling · Arnaud Doucet · Christophe Andrieu · Sebastian J. Vollmer · Pierre Jacob -
2015 Poster: MCMC for Variationally Sparse Gaussian Processes »
James Hensman · Alexander Matthews · Maurizio Filippone · Zoubin Ghahramani -
2015 Poster: Expectation Particle Belief Propagation »
Thibaut Lienart · Yee Whye Teh · Arnaud Doucet -
2014 Poster: Asynchronous Anytime Sequential Monte Carlo »
Brooks Paige · Frank Wood · Arnaud Doucet · Yee Whye Teh -
2014 Oral: Asynchronous Anytime Sequential Monte Carlo »
Brooks Paige · Frank Wood · Arnaud Doucet · Yee Whye Teh -
2009 Poster: Bayesian Nonparametric Models on Decomposable Graphs »
Francois Caron · Arnaud Doucet -
2009 Tutorial: Sequential Monte-Carlo Methods »
Arnaud Doucet · Nando de Freitas -
2007 Spotlight: Bayesian Policy Learning with Trans-Dimensional MCMC »
Matthew Hoffman · Arnaud Doucet · Nando de Freitas · Ajay Jasra -
2007 Poster: Bayesian Policy Learning with Trans-Dimensional MCMC »
Matthew Hoffman · Arnaud Doucet · Nando de Freitas · Ajay Jasra