420-500: Final Exam Review• Final exam: 11/18 (Tue) 12:45pm-2:00pm, in HRBB 113.• Final exam material: slide04 (page 46 and beyond), slide05 (all),slide06 (all).• This is a closed book exam, however, see the next item.• You may bring 1 sheet of notes (US letter), however small thefonts may be (obviously it should be legible to you). You may useboth sides.• You may bring a calculator, although you may not need it at all.1Key Points: 1• First-order logic• Prenex normal form, skolemization• Substitution• Unifiers• Resolution: two-pointer method, efficiency• Question answering2Key Points: 2• Uncertainty• Decision theory example: how prob theory and decision theoryare combined• Probability basics: terminology, notations.• Joint probability distribution: concept• Conditional probability: definition, various ways of representingconditional prob.• Axioms of probability: basic axioms, and using them to provesimple equalities.• Bayes rule: definition and application.3Key Points: 3• Why and when is Bayesian analysis useful?– Disease example– Vision example• How to calculate priors from conditional distributions?• How is subjective belief utilized in Bayesian analysis?• What is the role of priors in Bayesian analysis?4Key Points: 4• How is subjective belief utilized in Bayesian analysis?• Bayesian updating: why does that make probabilistic inferenceefficient when multiple evidence comes in?• Belief network: definition, semantics, extracting probabilities ofcertain conjunction of events.5Key Points: 5• Constructing a belief network: what is the procedure? why doesnode ordering matter? how to order the nodes?• Inference in belief networks: what are the kinds of inference?what is the general method? (causal, evidential, etc.)6Key Points: 6• Types of learning• Inductive learning (concept)• Decision tree learning:– What is the embodied principle (or bias)?– How to choose the best attribute? Given a set of examples,choose the best attribute to test first.– What are the issues? noise, overfitting, etc.– Relationship between probability, degree of sur prise, degreeof uncertainty, entropy, and information (gain).– Know how to calculate entropy and info gain.7Key Points: 7• Neural networks: basics• The central nervous system: how it differs from conventional computers (noneed to memorize numbers, anatomical names).• Basic mechanism of synaptic information transfer (no need to memorizechemicals etc.)• Types of neural networks: two ways of classifying, by feedback and bytopology.• Perceptrons: basic idea, and the geometric interpretation. What is thelimitation? How to train?• Backprop: how does it overcome perceptrons, learning algorithm (basicidea).8Key Points: 8• Unsupervised learning in general– relationship between redundancy, structure/organization,channel capacity vs. information content– what can kind of things can unsupervised learning do?• SOM: basic learning rule• SOM: error measures• SOM applications9Key Points: 9• Recurrent networks and their uses• GA and