Week 3 (gaussians: Uni, Multi, Gamma) Flashcards
(22 cards)
4 levels of inference
Relevant of parametric distributions
MLE for Gaussian univariate
Bernoulli dist
Binomial distribution
MLE for Bernoulli
Overfitting on Bernoulli
1) if observed values are {0, 0, 0} , μ = 0
2) to avoid overfitting, use a prior
3) conjugate prior
Beta distribution
Bayesian Bernoulli
Asymptotic properties of Bayesian Bernoulli posterior
Predictive posterior for Bayesian Bernoulli
Multivariate Gaussian pdf
Geometry of Multivariate Gaussian
1st moment of Multivariate Gaussian
2nd moment of Multivariate Gaussian and covariance
Bayes theorem for Multivariate Gaussian
Construct log likelihood for multi variate Gaussian
MLE for multi variate Gaussian
Analysis of MLE for multi variate Gaussian
Where this alternate MLE for Σ is unbiased
Bayesian inference for univariate Gaussian
The Gaussian-gamma prior enables conjugate updates
Gamma distribution
CLT
