Implementation of Prior, Rejection, Likelihood and Gibbs Sampling
-
Updated
Feb 4, 2024 - Python
Implementation of Prior, Rejection, Likelihood and Gibbs Sampling
Code repository for the paper No-Regret Approximate Inference via Bayesian Optimisation, published at UAI 2021
This project implements both exact and approximate inference techniques for Bayesian Networks using enumeration and rejection sampling, respectively. It processes Bayesian Network structures in XMLBIF format, accepting command-line inputs to compute the posterior distribution of a query variable given observed evidence.
Expectation Maximisation, Variational Bayes, ARD, Loopy Belief Propagation, Gaussian Process Regression
FAIKR MOD3 project
Code repository for the UAI 2020 paper "Active learning of conditional mean embeddings via Bayesian optimisation" by S. R. Chowdhury, R. Oliveira and F. Ramos.
My undergraduate honours project, with others' private information/code removed.
Codes for 'Improving Hyperparameter Learning under Approximate Inference in Gaussian Process Models' (ICML 2023)
Correcting predictions for approximate Bayesian inference
Denoise a given image using Loopy Belief Propagation
STOT: Single-Target Object Tracking using particle and Kalman filters [with a bonus multi-target].
Simulation-based inference using SSNL
Probabilistic approach to neural nets - modern scalable approximate inference methods
Empirical analysis of recent stochastic gradient methods for approximate inference in Bayesian deep learning, including SWA-Gaussian, MultiSWAG, and deep ensembles. See report_localglobal.pdf.
An implementation of loopy belief propagation for binary image denoising. Both sequential and parallel updates are implemented.
Approximate Ridge Linear Mixed Models (arLMM)
Variational Bayesian decision-making for continuous utilities
Benchmark of posterior and model inference algorithms for (moderately) expensive likelihoods.
Add a description, image, and links to the approximate-inference topic page so that developers can more easily learn about it.
To associate your repository with the approximate-inference topic, visit your repo's landing page and select "manage topics."