Bayesian Models
A suite of high-performance Python implementations for Gaussian Process Regression, Finite Mixture Models, and Hidden Markov Models.
This project consists of highly optimized Python implementations of three fundamental Bayesian statistical models. Each implementation focuses on computational efficiency, numerical stability, and vectorization, replacing standard iterative loops with linear algebra operations (NumPy/SciPy) and JIT compilation (Numba) where appropriate.
Content
GP.py Model: Gaussian Process Regression Key Techniques: Cholesky Decomposition, SciPy cdist
BayesianFM.py Model: 3D Bayesian Finite Mixture Key Techniques: Vectorized Gibbs Sampling, Log-Sum-Exp Trick
BayesianHMMGibbs.py Model: Bayesian Hidden Markov Model Key Techniques: FFBS Algorithm, Numba JIT Compilation