On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

169,341 papers · 148 categories

Trend · papers per month

count % of corpus

5 results for “Subposteriors”

Merging MCMC Subposteriors through Gaussian-Process Approximationsstat.CO

New method combines MCMC subposteriors using Gaussian processes.

problem Scaling MCMC for large datasets.

method Gaussian-process approximation for subposteriors, recombining with importance sampling.

result Approximation of full posterior with improved efficiency.

MCMC Bayesian inference Gaussian process Subposteriors Hamiltonian Monte Carlo

2016-05-27abs ↗pdf ↗

Embarrassingly parallel MCMC using deep invertible transformationscs.LG

A new method uses deep invertible transformations to parallelize MCMC for large datasets.

problem Scaling MCMC methods to large distributed datasets.

method Introduces a deep invertible transformation to approximate subposteriors, enabling efficient parallel MCMC.

result Demonstrates superior performance compared to existing methods in various challenging scenarios.

MCMC parallel computing Bayesian inference deep learning invertible transformations

2019-03-11abs ↗pdf ↗

Parallel MCMC Without Embarrassing Failuresstat.ML

A new method combines MCMC results to avoid failures in parallel computing.

problem Parallel MCMC's sensitivity to subposterior sampling issues leads to failures.

method Parallel Active Inference (PAI) uses Gaussian Process (GP) surrogate modeling and active learning.

result PAI successfully combines MCMC results where previous methods fail.

MCMC Parallel Computing Gaussian Process Active Learning Bayesian Inference

2022-02-22abs ↗pdf ↗

Parallelising MCMC via Random Forestsstat.CO

The paper parallelizes MCMC using random forests for big data Bayesian computation.

problem Efficiently computing Bayesian models in large datasets.

method Embed random forests into divide-and-conquer MCMC framework, using scaled subposteriors as proposal distributions.

result The approach works well across various model types, including misspecified models.

MCMC Bayesian computation Random forests Big data Importance sampling

2019-11-21abs ↗pdf ↗

Diffusion Generative Modelling for Divide-and-Conquer MCMCstat.ML

New method merges MCMC samples without distributional assumptions.

problem Efficiently merging MCMC samples from disjoint subsets.

method Diffusion generative modelling for density approximation.

result Outperforms existing methods on high-dimensional problems.

MCMC Density estimation Parallel computing Posterior merging

2024-06-17abs ↗pdf ↗