Bayesian Data Analysis of Nonparametric Models in Clojure
summary
A Duke team formulates a Bayesian nonparametric model for spike train intensity and implements Gibbs and Metropolis MCMC in Clojure using Apache Commons math and a custom distributions library (MichaelLinden/distributions) to sample posteriors, posterior predictives and Gaussian process functions. The pipeline recovers true simulated curves, quantifies posterior probability of switching versus non switching, and reports evidence of neural multiplexing in dual‑sound trials, showing the code and inference actually work.
AI-written summary · about
chapters (16)
- 0:00Start
- 0:57Introduction to Bayesian Statistics
- 1:45What Is Closure
- 6:14What Is Bayesian Inference
- 6:31Bayes Rule
- 8:55Model Using Sparse Regression
- 10:03Markov Chain Monte Carlo Algorithms
- 14:47Examples
- 15:47Truncated Distributions
- 16:03Mixture Distributions
- 16:43Posterior Distribution
- 17:35Posterior Predictive Distribution
- 17:57Sampling Algorithms Used for Sampling Non-Standard Densities
- 18:27Nonparametric Regression
- 20:22Gaussian Processes
- 24:54Gibbs Sampler
related talks
25mScaling Bayesian Experimentation
34mIntro to Probabilistic Programming and Clojure’s Anglican
32mBrain Data in Clojure
24mThe Return of Clojure Data Science
40m"Clojure in the Fintech Ecosystem"
33mClojure Data Science
41mInteractive GPU programming with ClojureCUDA & ClojureCL
26mDoing data science with Clojure: the ugly, the sad, the joyful