-
Notifications
You must be signed in to change notification settings - Fork 2
Home
Yee Whye Teh edited this page May 17, 2016
·
27 revisions
Weekly Reading Group on Bayesian Machine Learning
- When? Every Tuesday at 2:30pm
- Where? LG.04, Department of Statistics, 24-29 St Giles'
- Administrators: [email protected] or [email protected]
| Date | Topic | Presenters |
|---|---|---|
| 26 January 2016 | External talk - "Extremal point process of the branching Brownian motion" | Julien Berestycki |
| 2 February 2016 | W&J chap 3 - 4.1 | Leonard Hasenclever + Stefan Webb |
| 9 February 2016 | W&J rest of chap 4, (chap 5) | Thibaut Lienart, Xiaoyu Lu |
| 16 February 2016 | Leave Pima Indians Alone | Marco, Valerio |
| 23 February 2016 | Convergent EP for Dynamic Bayes Nets (Heskes & Zoeter), extended version | |
| 1 March 2016 | visitor talk | Kamalika Chaudhuri |
| 8 March 2016 | Stochastic EP, BB-alpha |
References to Literature SDE-SGMCMC-Literature>
| Date | Topic | Presenters |
|---|---|---|
| 19 April | SDEs | Tigran Nagapetyan Duncan, Computational Stochastic Processes p.59-61 and sections 4.5-4.8 without 4.7.1 (please read before Tuesday). Also consider chapter 3 of [Omiros notes] (http://www.econ.upf.edu/~omiros/course_notes.pdf) for very gentle introduction , Recommended Reading: |
| 26 April | SDEs and their discretisations | Tigran Nagapetyan and Sebastian Vollmer |
| 03 May | Ergodicity of SDEs and their discretisation | Sebastian Vollmer |
Possible Papers to discuss:
Variational inference and expectation propagation (Yee Whye)
- Wainwright & Jordan
- Leave Pima Indians Alone (Ridgeway & Chopin)
- EP for approximate inference (Heskes & Zoeter)
- BB-alpha
- Convergent EP (Lobato),
- Fast convergent algorithms for EP (Seeger, Nickish)
- EP for GP (which paper?)
- Variational inference for Monte Carlo objectives
SDEs and SGMCMC (Sebastian)
- Emily Fox Family of SDEs paper
- Changyou Chen & Lawrence Carin series (Dec 2015)
Bayesian Nonparametrics and stochastic processes
- [Gnedin & Pitman. Notes on the occupancy problem with infinitely many boxes: general asymptotics and power laws. Proba. surveys.]
- Harry Crane on Ewens' Sampling Formula
- Asmussen & Rosinsky, Cohen & Rosinsky on Gaussian tail approximations of Levy processes
- Tree processes (DDT, PYDT, coalescent, Gibbs fragmentation, continuum random trees?)
- Harry Crane on networks
- Neil Sheppard on Levy processes
- GP inducing points
- GP kernel structure learning (Duvenaud et al arxiv:1302.4922v4)
- Structure learning in matrix factorisation (Grosse et al)
Bayesian learning in general
- Bayesian dark Knowledge (Korattikara et al)
External speakers
- Andriy Mnih, Nicholas Heess, Charles Blundell
- Richard Xu
Model checking, evaluation (BDA Gelman)
- Posterior predictive checks
- WAIC