SusanHolmes
summary
Claims a reproducible pipeline that quantifies uncertainty in manifold-style projections and rankings, using Bayesian uni-dimensional scaling with heteroscedastic variance and Mahalanobis-style rescaling, implemented in Stan (HMC and variational Bayes tried). Builds a registration step — registration via RV-like matrix correlation — to align posterior MDS/PCA samples, produces contour credibility regions, and ships code and apps (GitHub and Philo seek) that succeed at locating clusters and gradients in microbiome and imaging data.
AI-written summary · about
chapters (17)
- 0:00Start
- 0:59Uncertainties and Distances
- 3:38Self Segmentation
- 5:33Mahalanobis Distance
- 8:48Variables Are Vectors in the Space of Points
- 12:04Linear Registration
- 12:40Correlation between Two Distance Matrices
- 13:44Multi-Dimensional Scaling
- 17:293d Mapping of the House of Representatives
- 20:27T-Sne
- 22:47The Uncertainty of the Projection of the Points
- 27:59Bayesian Nonparametric Model
- 31:03Brain Imaging
- 31:38Registration
- 35:22Studies of the Vaginal Microbiome
- 39:20Metric Learning
- 39:53Measures of Uncertainty
related talks
34mFast, beautiful and easy Bayesian modelling: Can you have it all?
38mManifold Learning and Dimensionality Reduction for Data Visualization
47mScale Calibration for High-Dimensional Robust Regression
51mRandom Projections for Probabilistic Inference
26mAnthony Bak, Principal Data Scientist and Mathematician, Ayasdi
18mMulti-Agent Systems for Traceable Bayesian Workflow
01h 03mAn Overview of Probabilistic Programming
29mMotivating Probabilistic Programming