But what is a Gaussian process? Regression while knowing how certain you are
summary
Defines a Gaussian process as a predictive distribution that returns a mean and variance not a single point estimate and derives posteriors by conditioning a multivariate normal, implements examples in Python using GPy, uses an RBF kernel so covariance falls with distance, shows a three‑point regression where 95% intervals widen when extrapolating and narrow when correlated observations arrive, demonstrates Bayesian optimization and a neural‑net+GP hybrid on housing price data, and reports concrete numerical interval changes.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 1:03Motivating regression example
- 3:09Limitations of predictions
- 4:33Gaussian processes and uncertainty
- 6:59Housing price analogy
- 9:10Analogy to machine learning
- 11:16Univariate Gaussian basics
- 13:04Multivariate Gaussians
- 17:03Updating with new data
- 21:57Three variables and housing prices
- 25:07Modeling 20 house prices
- 26:53From finite to infinite dimensions
- 28:59GPs as infinite Gaussians
- 29:21Implementing GPs in Python
- 30:24Closing remarks
related talks
01h 30mA Primer on Gaussian Processes for Regression Analysis
40mBdrygp: A Boundary-Integrated Gaussian Process Model For Computer Code Emulation
01h 30mMaking Gaussian Processes Useful
01h 33mMastering Gaussian Processes with PyMC
39mBayesian optimisation with scikit-learn
48mBayesian Optimization of Gaussian Processes with Applications to Performance Tuning
49mSo you think you know about linear regression ...
44mAre You Sure about That?! Uncertainty Quantification in AI