Publications (15)
Accelerated Parallel Tempering via Neural Transports
Leo Zhang, Peter Potaptchik, Jiajun He +5
Markov Chain Monte Carlo (MCMC) algorithms are essential tools in computational statistics for sampling from unnormalised probability distributions, but can be fragile when targeti…
A Unified Framework for U-Net Design and Analysis
Christopher Williams, Fabian Falck, George Deligiannidis +3
U-Nets are a go-to, state-of-the-art neural architecture across numerous tasks for continuous signals on a square such as images and Partial Differential Equations (PDE), however t…
Score-Optimal Diffusion Schedules
Christopher Williams, Andrew Campbell, Arnaud Doucet +1
Denoising diffusion models (DDMs) offer a flexible framework for sampling from high dimensional data distributions. DDMs generate a path of probability distributions interpolating…
Parallel Tempering With a Variational Reference
Nikola Surjanovic, Saifuddin Syed, Alexandre Bouchard-Côté +1
Sampling from complex target distributions is a challenging task fundamental to Bayesian inference. Parallel tempering (PT) addresses this problem by constructing a Markov chain on…
Amortized Sampling with Transferable Normalizing Flows
Charlie B. Tan, Majdi Hassan, Leon Klein +5
Efficient equilibrium sampling of molecular conformations remains a core challenge in computational chemistry and statistical inference. Classical approaches such as molecular dyna…
Parallel Tempering on Optimized Paths
Saifuddin Syed, Vittorio Romaniello, Trevor Campbell +1
Parallel tempering (PT) is a class of Markov chain Monte Carlo algorithms that constructs a path of distributions annealing between a tractable reference and an intractable target,…
Uniform Ergodicity of Parallel Tempering With Efficient Local Exploration
Nikola Surjanovic, Saifuddin Syed, Alexandre Bouchard-Côté +1
Non-reversible parallel tempering (NRPT) is an effective algorithm for sampling from target distributions with complex geometry, such as those arising from posterior distributions…
Optimised Annealed Sequential Monte Carlo Samplers
Saifuddin Syed, Alexandre Bouchard-Côté, Kevin Chern +1
Annealed Sequential Monte Carlo (ASMC) samplers are special cases of SMC samplers where the sequence of distributions can be embedded in a smooth path of distributions. Using this…
Conditional Diffusion Sampling
Francisco M. Castro-MacÃas, Pablo Morales-Ãlvarez, Saifuddin Syed +3
Sampling from unnormalized multimodal distributions with limited density evaluations remains a fundamental challenge in machine learning and natural sciences. Successful approaches…
Non-Reversible Parallel Tempering: a Scalable Highly Parallel MCMC Scheme
Saifuddin Syed, Alexandre Bouchard-Côté, George Deligiannidis +1
Parallel tempering (PT) methods are a popular class of Markov chain Monte Carlo schemes used to sample complex high-dimensional probability distributions. They rely on a collection…
The Cosine Schedule is Fisher-Rao-Optimal for Masked Discrete Diffusion Models
Leo Zhang, Saifuddin Syed
In this work, we study the problem of choosing the discretisation schedule for sampling from masked discrete diffusion models in terms of the information geometry of the induced pr…
Local Exchangeability
Trevor Campbell, Saifuddin Syed, Chiao-Yu Yang +2
Exchangeability -- in which the distribution of an infinite sequence is invariant to reorderings of its elements -- implies the existence of a simple conditional independence struc…
Pigeons.jl: Distributed Sampling From Intractable Distributions
Nikola Surjanovic, Miguel Biron-Lattes, Paul Tiede +3
We introduce a software package, Pigeons.jl, that provides a way to leverage distributed computation to obtain samples from complicated probability distributions, such as multimoda…
CREPE: Controlling Diffusion with Replica Exchange
Jiajun He, Paul Jeha, Peter Potaptchik +5
Inference-time control of diffusion models aims to steer model outputs to satisfy new constraints without retraining. Previous approaches have mostly relied on heuristic guidance o…
autoMALA: Locally adaptive Metropolis-adjusted Langevin algorithm
Miguel Biron-Lattes, Nikola Surjanovic, Saifuddin Syed +2
Selecting the step size for the Metropolis-adjusted Langevin algorithm (MALA) is necessary in order to obtain satisfactory performance. However, finding an adequate step size for a…