Oral exam: list of questions
Aggregazione dei criteri
- Introduce the Empirical Risk Minimization principle
- Define PAC Learning
- Define VC Dimension(Vapnik–Chervonenkis)
- Introduce the basic ideas of Probabilistic Inference
- Introduce Bayesian Networks and their factorization
- Conditional Independence in Bayesian Networks
- Introduce Random Markov Fields
- Define Factor Graphs and How to convert Bayesian Networks and Markov Random Fields to Factor Graphs
- Describe the Sum Product algorithm
- Describe the Max Plus algorithm
- Define HMM and different Inference problems in HMM
- Discuss closure properties of the Gaussian Distribution
- How to transform a generic Gaussian Distribution to a Standard Gaussian via Principal components
- Introduce the Bayesian Estimation Principles
- Introduce Bayesian Linear Regression
- Prior and posterior over parameters in Bayesian Linear Regression
- Predictive distribution n Bayesian Linear Regression
- Discuss Model Evidence and hyper parameter optimization
- Introduce the idea of Effective number of parameters
- Discuss Bayesian Model Comparison
- Describe Laplace Approximation
- Discuss Bayesian Information Content for Model Comparison
- Introduce Bayesian Logistic Regression
- Rejection sampling
- Importance sampling
- Introduce Markov chain and the Detailed Balance condition
- Introduce Markov Chain Monte Carlo and the Metropolis Hastings criterion
- Discuss issues of vanilla MCMC
- Gibbs Sampling
- Discuss Convergence Diagnostics and the That index
- Introduce the ideas of Effective sample size
- Hamiltonian Monte Carlo
- Introduce the problem that can be solved by EM
- Introduce and derive the Evidence lower bound
- Discuss the Expectation Maximization algorithm (both E-step and M-step)
- Discuss convergence of the EM algorithm
- Discuss EM algorithm for Gaussian Mixtures
- Introduce the problem formulation for Variational Inference
- Introduce Mean Field Variational Inference
- Example of Mean Field Variational Inference on Gaussian distribution
- Variational Inference with direct and inverse KL
- Variational Linear Regression
- Introduce Black box Variational Inference
- Discuss how to compute the gradient of the ELBO for a non-reparameterizable variational distribution
- Discuss Rao-Blackwellization for variance reduction
- Discuss Control variates for variance reduction
- Discuss Bayesian Neural Networks
- Introduce the generative modelling problem
- Introduce Autoencoding Variational Bayes
- Introduce Denoising Diffusion Models
Ultime modifiche: sabato, 22 giugno 2024, 13:37