Home Library Explore SAT

Home Library Explore SAT

Bayesian inference #1 Flashcards | StudBud

Explore/Bayesian inference #1

Bayesian inference #1

5 terms·StudBud Administration·100 plays

Imported from external platform

Flashcards5 cardsFPractice quizPHost liveL

Questions 5

Answers

1
What is the correct version of the Bayes theorem:
Likelihood = PDF(data, parameters)
Bayesian statistics are always better than Maximum Likelihood
Posterior = (likelihood * prior)/ marginal likelihood
Marginal likelihood = posterior * likelihood of the data * prior
2
log-transform the following term: log( 1 / sqrt(a) )
1/a
42
-2 log(a)
-0.5 * log(a)
3
What is the probability of observing a bear and a panda?
The product between P(bear) and P(panda)
1
The sum of P(bear) and P(panda)
0
4
The y-value for a given x on a PDF is the:
Relative probability of the point
Same, no matter what parameters we choose
Extension of an adobe acrobat document
Absolute probability of the point
5
Why do we log-transform the probability density functions?
A computer can't distinguish between a very small number and 0
It simply looks fancier and more "mathy"
Bayesian models can only deal with log-likelihoods
Working with log-likelihoods speeds up computation significantly