Lecture 2: Statistical learning primer for biologists Alan Qi Purdue Statistics and CS Jan. 15, 2009

Outline Basics for probability Regression Graphical models: Bayesian networks and Markov random fields Unsupervised learning: K-means and Expectation maximization

Probability Theory Sum Rule Product Rule

The Rules of Probability Sum Rule Product Rule

Bayes’ Theorem posterior  likelihood × prior

Probability Density & Cumulative Distribution Functions

Expectations Conditional Expectation (discrete) Approximate Expectation (discrete and continuous)

Variances and Covariances

The Gaussian Distribution

Gaussian Mean and Variance

The Multivariate Gaussian

Gaussian Parameter Estimation Likelihood function

Maximum (Log) Likelihood

Properties of and Unbiased Biased

Curve Fitting Re-visited

Maximum Likelihood Determine by minimizing sum-of-squares error, .

Predictive Distribution

MAP: A Step towards Bayes Determine by minimizing regularized sum-of-squares error, .

Bayesian Curve Fitting

Bayesian Networks Directed Acyclic Graph (DAG)

Bayesian Networks General Factorization

Generative Models Causal process for generating images

Discrete Variables (1) General joint distribution: K 2 -1 parameters Independent joint distribution: 2(K-1) parameters

Discrete Variables (2) General joint distribution over M variables: KM -1 parameters M node Markov chain: K-1+(M-1)K(K-1) parameters

Discrete Variables: Bayesian Parameters (1)

Discrete Variables: Bayesian Parameters (2) Shared prior

Parameterized Conditional Distributions If are discrete, K-state variables, in general has O(K M) parameters. The parameterized form requires only M + 1 parameters

Conditional Independence a is independent of b given c Equivalently Notation

Conditional Independence: Example 1

Conditional Independence: Example 1

Conditional Independence: Example 2

Conditional Independence: Example 2

Conditional Independence: Example 3 Note: this is the opposite of Example 1, with c unobserved.

Conditional Independence: Example 3 Note: this is the opposite of Example 1, with c observed.

“Am I out of fuel?” B = Battery (0=flat, 1=fully charged) And hence B = Battery (0=flat, 1=fully charged) F = Fuel Tank (0=empty, 1=full) G = Fuel Gauge Reading (0=empty, 1=full)

“Am I out of fuel?” Probability of an empty tank increased by observing G = 0.

“Am I out of fuel?” Probability of an empty tank reduced by observing B = 0. This referred to as “explaining away”.

The Markov Blanket Factors independent of xi cancel between numerator and denominator.

Markov Random Fields Markov Blanket

Cliques and Maximal Cliques

Joint Distribution where is the potential over clique C and is the normalization coefficient; note: M K-state variables  KM terms in Z. Energies and the Boltzmann distribution

Illustration: Image De-Noising (1) Original Image Noisy Image

Illustration: Image De-Noising (2)

Illustration: Image De-Noising (3) Noisy Image Restored Image (ICM)

Converting Directed to Undirected Graphs (1)

Converting Directed to Undirected Graphs (2) Additional links: “marrying parents”, i.e., moralization

Directed vs. Undirected Graphs (2)

Inference on a Chain Computational time increases exponentially with N.

Inference on a Chain

Supervised Learning Supervised learning: learning with examples or labels, e.g., classification and regression Linear regression (the example we just given), Generalized linear models (e.g, probit classification), Support vector machines, Gaussian processes classifications, etc. Take CS590M-Machine Learning in 2009 fall.

Unsupervised Learning Supervised learning: learning with examples or labels, e.g., classification and regression Unsupervised learning: learning without examples or labels, e.g., clustering, mixture models, PCA, non-negative matrix factorization

K-means Clustering: Goal

Cost Function

Two Stage Updates

Optimizing Cluster Assignment

Optimizing Cluster Centers

Convergence of Iterative Updates

Example of K-Means Clustering

Mixture of Gaussians Mixture of Gaussians: Introduce latent variables: Marginal distribution:

Conditional Probability Responsibility that component k takes for explaining the observation.

Maximum Likelihood Maximize the log likelihood function

Maximum Likelihood Conditions (1) Setting the derivatives of to zero:

Maximum Likelihood Conditions (2) Setting the derivative of to zero:

Maximum Likelihood Conditions (3) Lagrange function: Setting its derivative to zero and use the normalization constraint, we obtain:

Expectation Maximization for Mixture Gaussians Although the previous conditions do not provide closed-form conditions, we can use them to construct iterative updates: E step: Compute responsibilities . M step: Compute new mean , variance , and mixing coefficients . Loop over E and M steps until the log likelihood stops to increase.

Example EM on the Old Faithful data set.

General EM Algorithm

EM as Lower Bounding Methods Goal: maximize Define: We have

Lower Bound is a functional of the distribution . Since and , is a lower bound of the log likelihood function .

Illustration of Lower Bound

Lower Bound Perspective of EM Expectation Step: Maximizing the functional lower bound over the distribution . Maximization Step: Maximizing the lower bound over the parameters .

Illustration of EM Updates