advertisement

CS 540-3 Introduction to Artificial Intelligence Spring 2015 CS 540-3 Final Exam Topics 1. Constraint Satisfaction Problem formulation in terms of variables, domains and constraints, constraint graph, depth-first search, backtracking with consistency checking, most constrained variable heuristic, most constraining variable heuristic, least constraining value heuristic, min-conflicts heuristic, minconflicts algorithm, forward checking algorithm, arc consistency algorithm (AC-3). 2. Neural Networks Perceptron, LTU, activation functions, bias input, input units, output units, Perceptron learning rule, Perceptron learning algorithm, epoch, weight space, input space, linearly separable, credit assignment problem, multi-layer feed-forward networks, hidden units, sigmoid function, backpropagation algorithm, gradient descent search in weight space. (Nothing on deep learning.) 3. Support Vector Machines Maximum margin, definition of margin, kernel trick, support vectors, slack variables. 4. Reasoning under Uncertainty Random variable, mutually exclusive, prior probability, 3 axioms of probability, joint probability, conditional probability, posterior probability, full joint probability distribution, degrees of freedom, summing out, marginalization, normalization, product rule, chain rule, conditionalized version of chain rule, Bayes’s rule, conditionalized version of Bayes’s rule, addition/conditioning rule, independence, conditional independence, naïve Bayes classifier. 5. Bayesian Networks Bayesian network DAG, conditional probability tables, space saving compared to full joint probability distribution table, conditional independence property defined by a Bayesian network, inference by enumeration from a Bayesian network, naïve Bayes classifier as a Bayesian network. 6. Speech Recognition Phones, phonemes, speech recognition using Bayes’s rule, language model, acoustic model, bigram model, trigram model, first-order Markov assumption, probabilistic finite state machine, first-order Markov model, state transition matrix, vector, computing conditional probabilities from a Markov model, hidden Markov model, observation likelihood matrix, computing joint probabilities and conditional probabilities from an HMM by enumeration. (Nothing on Forward algorithm, Viterbi algorithm, Forward-Backward algorithm, Siri, particle filters, tracking in video.) 7. Computer Vision Viola-Jones face detection algorithm, boosting ensemble learning, AdaBoost algorithm, Eigenfaces algorithm, nearest-neighbor classification, image space, face space, average face, eigenvalues, eigenvectors, dimensionality reduction. (Nothing on PCA or SVD.) 8. Propositional Logic Propositional logic, propositional symbol, interpretation, model, satisfiable, unsatisfiable, inconsistent, contradiction, valid, tautology, knowledge base, entailment, deductive inference, soundness, completeness, inference by enumeration in PL, natural deduction, sound inference rules, modus ponens, logical equivalence, monotonicity property, resolution refutation algorithm, resolution rule of inference, conjunctive normal form, clause, empty clause, Horn clause, literal, forward chaining, backward chaining. (Nothing on first-order logic.)