Exchangeability
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact Konstantina Palla.
Data is called exchangeable if the probability of a sample depends only on the values that are observed, but not on the order in which they occur. This simple assumption has a rather surprising consequence: An
exchangeable distribution can be decomposed completely into a shared, underlying random “pattern” and a component representing independent “randomness” in each observation. This result is de Finetti’s theorem. I will devote roughly the first half of the talk to the theorem and its immediate implications for Bayesian statistics. In the remaining time, I hope to provide a glimpse of more recent results and of the fascinating picture which emerges with them, which relates symmetry and invariance principles to the design of probability models.
This talk is part of the Machine Learning Reading Group @ CUED series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
|