Midterm Solved - Artificial Intelligence and Engineering Application | ECE 4524, Exams of Electrical and Electronics Engineering

Download Midterm Solved - Artificial Intelligence and Engineering Application | ECE 4524 and more Exams Electrical and Electronics Engineering in PDF only on Docsity! MIDTERM EXAMINATION ECE 4524: Artificial Intelligence and Engineering Applications Fall 2008 NAME (print): —_ SOLUTION S STUDENT ID: Honor system pledge: T have neither given nor received unauthorized aid on this exam. SIGNATURE: Ground rales. Please read carefully. Do noi open the exam until the instructor says to begin. This exam is closed-book. No books or notes are permitted. No electronic devices are permitted, except for a simple chronometer. The exam is worth 126 points. Time limit: 75 minutes. The exam consists of 11 problems on 8 pages. (There is also an extra-credit problem at the end.) Be sure that all pages are present in your copy. Print your name or initials on each page. Work ail problems on the pages provided. If you need extra space, continue on the back of the sheet. Show how you arrive at each answer, and clearly indicate your final answers. Partial credit is possible for most problems. If you do not have time to complete a problem, describe briefly how you would finish it. (Anything is almost always better than a blank answer.} Ifa problem seems ambiguous, give your reasons and state your assumptions clearly. All work on this exam must be your own. If you have questions during the exam, ask the instructor. Do not communicate with another student. Print your name or initials on cach page of this exam. This will help ensure that you receive proper credit even if the pages become separated. Check the front of the classroom occasionally for important corrections or clarifications. Page 1 Your name or initials: Problem 1. (13 points) Indicate True or False for each of these statements. jf »)_F a a) Ff. a 7 _T 8) T Hh T pF pF »_F yp T Alis primarily concerned with developing fast techniques to play games such as chess. The name of the Python programming language reflects the snake-like features of the language. The Python language supports recursion. Depth-first search will always find a goal state, if one exists. Hill-climbing search is a form of irrevocable search. To order to avoid iocal maxima or minima in the search objective function, some search techniques use probability-based rules to select new states. On-line search refers to search problems for which an agent interleaves computation and action Deductive inference is truth-preserving. Inductive inference is truth-preserving. Now that a chess-playing machine has beaten the human world champion, it is reasonable to expect all remaining AI problems to be solved by the year 2020. Alpha-beta pruning reduces the complexity of minimax search from exponential to linear. Adversarial search can suffer from the positive horizon effect when short-term gains mask unavoidable consequences. The inference rule known as modus ponens can be expressed as [P A (P = QW > Q. Problem 2. (5 points) In the space below, give an example in English of an inductive inference. Do not use the same example that was given in class. Only 2 or 3 sentences should be sufiicient. Pay Jov's doa heard am bell before man) meals, Pavlov 3 doo bhecebre expec bed ame! Aa evens tome he Atord @ bel}, Page 2 Your name or initials: Problem 6. (5 points) Suppose that Joan wants to use A* search to solve the 8-puzzle. She is trying to select a good cost function of the form fn) = g(a) + A(n). She is currently considering these choices: g(n) = number of tile moves that have been made from the initial puzzle state to state n. h(n) = number of squares for state n that have contents different from the goal state. t! due would b = th: wing situation: For example, the value would be #(77) = 4 for the following situation: shade x i state na goal state sit iz t 2) {-jif2 sia, 3/415 3/4) 5 16 We (o[s]7} e{zisl — Is this a good cost function for Joan to use for A* search? Answer yes or no, and give a brief but convincing argument to support your conclusion. ; \ Kk which is phe o be ah Samehiongs gv eres tim aber Ax “ ee \ are of moves to reach the goal shat, This 3 hose for Stade x tipave < hve bu? Az, Jaan s suggestion dees nol smplemant A search, Problem 7. (12 points) State 2 brief and relevant facts about each of the following topics: a) Turing test 1 This was proposed in }9S50 as a best for hopelligence, 2 A maching tomraunitaes via text and peits To Convinga A humen tha? phe maching 13 human, b) Type A systems . Chava dtri2ed by massive amounts of Stach nying fast, siople SEFS. 2 Has bean sncee stu) for games Such as chess, c) Crossover (as used in genetic algorithms, 1. Sejeut new fonda) stahes by Stpacejing oe re-combining pach of pacent skates, 2 Tntreduces an ja ocpant random elemray wy phe search faced in on Aiftenp? po Ave local PR AKTTA Page 5 Your name or initials: Problem 8. (9 points} Consider the propositional logic statement —C — (D v B). a} The converse of this statement is b) If the original statement above is true, then its converse must also be true. (Irue orfats 7 (ve) > C c) The contrapositive of this statement is d) Ef the original statement above is true, then its contrapositive must also be true. (eeu False) e) Convert the statement —C => (D v E) to clause form. “boy v (DYE) IC v DY «E\ Problem 9. (12 points) Consider the following propositional logic axioms. Use resolution theorem proving (by refutation) to attempt to show that the Dy. aA A Le is true. Clearly state B)z the proof succeeds. Hee eis ene possib =(7A A RB #9 y 9 Axioms: nwe A ae 4) RALSE 3 BvD = KL 3) 8B 4) =D — ) > 8) A — wee the pot neg 5) Av7B phoeetece succeeds, eoctrn Page 6 Your name or initials: Problem 10. (12 poimts) Consider the following map-coloring problem. The diagram below represents 5 geographic regions, and each region must be assigned a color from the set {red, green, blue} such that no two adjacent regions receive the same color. E Cc a) In the space above and to the right, draw a consiraini_graph that represents this as a Constraint Satisfaction Problem (CSP). b) Consider the first few steps of a map-coloring search procedure that uses forward checking. (For this problem, do not consider additional constraint propagation steps.) The diagram below represents an initial assignment of possible values to each geographic region, where r, g, and b represent red, green, and blue, respectively. A B c D E rgb | rgb reb rgb rgb Assume that region B is now arbitrarily assigned the color red, as indicated in the table below. In the empty cells of the table below, show the possible values that remain for each region after the forward checking step. A B c D EL 9 b.. i ab rob | rab vi G c) Based on your answer to part (b), which geographic region(s) should be considered next for value assignment according to the “most constrained variable” heuristic? Ao Cy becanst tack has only 2 /abe)s remaining, d) Continue this sg4rch problem based on your answers above. Select a region that you indicated in part (c), and assign afvalue to it. Apply forward checking. In the table below, indicate the possible values that remain for eacH region. sos coven an Pee nrevameaseenn __P E a & th A, y bs ebitraci! | Wied label % phase chosces cause Phe Soe waved Checking. rocedure to reenove ‘9’ from Pr des C, O, vf. Page 7