CS430 Second Midterm Study Guide

• Chapter 4.1 (A*)
• Chapter 13 (all)
• Chapter 14.1, 14.2, 14.4
• Chapter 15.1, 15.2, 15.3, 15.5, 15.6
• Chapter 20.1, 20.2, 20.3
• Chapter 22 (all)
• Chapter 23 (all)

Specific Topics

• You should understand A* very well and be able to simulate it.
• You should understand probability: joint distributions, conditional probability, independence, conditional independence. You should expect questions in which you are given a joint distribution and asked to answer queries from it.
• You should understand Bayesian networks and how they relate the joint distribution. You should expect questions in which you are given some evidence and asked to execute the Variable Elimination Algorithm to infer the marginal probability of some variables.
• You should understand learning for Bayesian networks including the Laplace correction ("add-one smoothing"). You should expect questions in which you must compute a conditional probability table given a set of training examples.
• You should understand the EM algorithm as applied to gaussians and to ordinary Bayesian networks (i.e., with boolean random variables).
• You should be able to explain how a speech recognition system works (what are phones? what is a language model? etc.)
• You should be able to discuss the different phases of natural language understanding. What kinds of representations are used in each phase? What kinds of knowledge are needed in each phase? You should know what a "Speech Act" is and to be able to give some examples. You should be able to construct a parse tree for a sentence given a simple context-free grammar and a lexicon. What makes understanding natural languages hard?
• You should be able to discuss different applications of natural language processing: information retrieval, information extraction, and machine translation.