Efficient Bayesian inference with Hamiltonian Monte Carlo -- Michael Betancourt (Part 1)
Channel:MLSS Iceland 2014 · Length:1:29:42 · Published:2014-11-17 · Views:47971
Summary
Michael Betancourt's lecture at MLSS Iceland 2014 on efficient Bayesian inference with Hamiltonian Monte Carlo. It covers MCMC concepts and motivation and the problems of sampling in high-dimensional distributions.
Why HN recommends it
HN people return to it as a clear primer on MCMC and HMC and on why high dimensions behave counterintuitively.
it was a fantastic primer on the basic concepts and motivations of mcmc, and later hmc — HN, 2022
Data science & statistics · University lectures & courses
Subtitles (excerpt)
we've had some really nice talks about Bayesian inference about MCMC I'm gonna talked about some big data stuff what I want to get into today is when we take these MCMC methods when we take these kind of generic things that work okay in one two three dimensions and you try to push them to high dimensional systems the big systems that were you know kind of interested in applied statistics these days a lot of stuff starts falling apart and you have to be really really careful to make things that work so what we're interested in is building up kind of principled mathematical algorithms that by construction will work in these high dimensional spaces so what I'd like to do today is hopefully motivate why we need these very specific kind of algorithms and show you a lot of pathologies that arise if you're not careful so there'll be some math but not a lot of math mostly just ideas all right so hopefully it should be nice and interpretive all and then that motivation will bring us into Hamiltonian Monte Carlo which is a really really nice elegant framework for doing these kinds of structures and that's going to bring us into stand which is our implementation of HMC combined with a lot of very nice user facing interfaces and
On Hacker News
Recently watched this on Hamiltonian Monte Carlo, and it was a fantastic primer on the basic concepts and motivations of mcmc, and later hmc [VIDEO] — leecarraher, 2022-10
Not an expert in this field, but I'm guessing this is related to unintuitive nature of geometry in high-dimensional spaces. One rough example I can think of is the fact that the number of ways to move away from the origin increases exponentially (or even faster?) as the dimensionality goes up. There is _way_ more volume away from the origin than near it, I've seen this explained as something like… — euW3EeBe, 2024-05
This is an interesting take and kudos to the author for relaying a helpful way to think about high dimensional distributions. I really like and often come back to this talk by Michael Betancourt where the theme is quite similar: [VIDEO] — srs70187, 2017-11
Related videos
- How to Speak · MIT OpenCourseWare · 1:03:42
- 1. Introduction to Human Behavioral Biology · Stanford · 57:14
- LEADERSHIP LAB: The Craft of Writing Effectively · UChicago Social Sciences · 1:21:51
- A Better Default Colormap for Matplotlib | SciPy 2015 | Nathaniel Smith and Stéfan van der Walt · Enthought · 19:08
- MIT's Pathway to Fusion Energy (IAP 2017) - Zach Hartwig · MIT Plasma Science and Fusion Center · 1:11:39
- Stanford's Sapolsky On Depression in U.S. (Full Lecture) · Stanford · 52:28
- Justice: What's The Right Thing To Do? Episode 01 "THE MORAL SIDE OF MURDER" · Harvard University · 54:56
- Lec 1 | MIT 18.06 Linear Algebra, Spring 2005 · MIT OpenCourseWare · 39:49
- Randy Pausch's Last Lecture: Achieving Your Childhood Dreams · Carnegie Mellon University · 1:16:27
- Lecture 1A: Overview and Introduction to Lisp · MIT OpenCourseWare · 1:12:55
- Lecture 1A | MIT 6.001 Structure and Interpretation, 1986 · MIT OpenCourseWare · 1:12:55
- The Most IMPORTANT Video You'll Ever See (part 1 of 8) · wonderingmind42 · 9:17
- Lecture 1 | Machine Learning (Stanford) · Stanford · 1:08:40
- Confessions of a Converted Lecturer: Eric Mazur · UMBCtube · 1:20:08
- The Coming Collapse of the Middle Class with Elizabeth Warren · University of California Television (UCTV) · 57:37
- RL Course by David Silver - Lecture 1: Introduction to Reinforcement Learning · Google DeepMind · 1:28:12