SaaaS - Sampling as an Algorithm Service
summary
Explains that sampling recovers posteriors by building a naked for loop MCMC sampler in Python, then swaps to PyMC3 for speed and applies it to a Santa Kaggle bag-weight inference and a time-series switch point detection. Shows histograms and chains, reports convergence behavior (easy for small problems, uncertain for underdetermined bags), and compares to a TensorFlow gradient search that often finds a maximum but can get stuck where sampling gives uncertainty and posterior shape.
AI-written summary · about
chapters (10)
- 0:00Sampling as a service
- 1:04Direct vs indirect sampling
- 3:34Sampling normal parameters
- 8:53Python sampling algorithm
- 12:51Sampling benefits and uses
- 16:20Kaggle gift weights example
- 21:38Switch points in time series
- 27:11Sampling vs gradient descent
- 30:22When to use sampling
- 30:43Q&A
related talks
31mChainsai:- facilitating sampling of multimodal probability distributions
44mAn introduction to PyMC3
01h 31mProbabilistic Programming and Bayesian Inference in Python
01h 27mBayesian Time Series
57mA Quickstart Guide to PyMC3
39mProbablistic Programming Data Science with PyMC3
01h 27mProbabilistic Python: An Introduction to Bayesian Modeling with PyMC
42mBayesian Non-parametric Models for Data Science using PyMC3