perm filename H9[224,DBL] blob
sn#104024 filedate 1974-05-29 generic text, type T, neo UTF8
00100 CS224 Handout #9 Thursday, May 30, 1974
00200
00300 1. There will be a special CS224 guided tour of the AI lab
00400 this Monday evening, June 3, at 6:30 PM. The director of the
00500 lab, and some researchers, will conduct you through the
00600 building. Highlights: See the hand drop something;
00700 Talk to PARRY (Colby's system) and see if you can get
00800 it to punch you in the nose; Observe some of the people you've
00900 read about in their natural habitat.
01000 If you need a ride, or can offer someone a ride, stop by at the
01100 end of this class. The AI Lab is gotten to as follows: take Page
01200 Mill road West; just after passing Route 280, you come to
01300 Arastrodero Road; make a right and continue on for a few
01400 minutes; the "DC Power Lab Building" will be up on a hill on your
01500 right. Turn right, go up the steep hill, and the AI lab is the
01600 second big building you pass on your left (lots of cars parked
01700 in front).
01800
01900
02000 2. Administrative
02100 This homework set is the last one. If you've skipped any assignments,
02200 you are in danger of receiving a poor grade; we strongly suggest that
02300 you spend a few minutes and try (and turn in) each homework. There
02400 is no penalty for lateness. Here is a class list; an asterisk
02500 denotes that the corresponding homework has been turned in,
02600 graded, and returned. Check these and report any disagreements
02700 quickly (this is correct up to May 27, and doesnt include
02800 homeworks turned in later than that date). Also check the column
02900 indicating that you are taking CS224 for a grade(G), for
03000 Pass/No Credit (P), or Auditing it (A).
03100
03200 All homework must be in by 5 PM, Monday, June 10, 1974. No exceptions!
03300
03400 Although there is no class Thursday, June 6, the TA will have
03500 (final) office hours that day, from noon to 3 PM, in Polya 263.
03600
03700 Class List, CS 224, Spring, 1974.
03800
03900
04000 NAME G/P/A YEAR HW1 HW2 HW3 HW4 HW5 HW6 HW7 HW8
04100 Anderson, Michael G * ** ** * ** * **
04200 Bahar, Jesu Philippe ? * ** * * *
04300 Baron, A. Chris G * ** * * **
04400 Barstow, Dave A *
04500 Barthelow, Mivhael G *
04600 Baudoin, Claude G * ** * * *
04700 Bentley, Jon L. P * * * *
04800 Berger, Steven G * * ** * ** **
04900 Bettendorff, Roland G * ** * * * **
05000 Bizzarri, Maurice G * * *
05100 Budney, Robert A *
05200 Campbell, Olin P * ** * * * * *
05300 Capitant, P. A *
05400 Carhart, Herbert G * *
05500 Chan, Tony P * * * *
05600 Chirieleison, Dominic G * ** * * * * **
05700 Cohn, Avra G * ** ** * ** ** **
05800 Cordell, Warren N. G * *
05900 DeMers, David G * * * * *
06000 Dixon, Peter G * * * *
06100 Dossi, Melissa Ann G *
06200 Dourson, Robert A *
06300 Drazovich, Bob G * ** ** * ** **
06400 Drsydale , Robert Scot G * * ** ** ** * *
06500 Flammer, George G * ** * ** *
06600 Fong, Paul G * ** *
06700 Ginsparg, Jerrold G * ** * * * *
06800 Glower, Rodger M. G * * ** * ** * **
06900 Guttman, Tony G *
07000 Harvey, Brian G * *
07100 Hole, Bonnie G * **
07200 Johnson, C. Richard A *
07300 Kagan, ? G * ** * * *
07400 Kant, Elaine G * ** ** * ** * **
07500 Karlin, Martin G * * *
07600 Kato, Hisatoyo A *
07700 Kim, Scott * *
07800 King, Jonathan J. ** ** * * *
07900 Kotaro, Hirasawa A *
08000 Larsen, Inge G * * * *
08100 Lawson, Donna J. G * * * * * *
08200 Lee, Ruby G * * ** **
08300 Lennert, David G * ** ** * * * *
08400 Levinthal, Elliott G *
08500 Levinthal, Michael G *
08600 Malloy, Thomas G * * * * ** **
08700 Meyer, Bertrand P. G * **
08800 Miller, Thomas Robert G * *
08900 Mitchell, Tom A *
09000 Muir, Donald E. ? * * * * * * *
09100 Northlich, W.R. G *
09200 Palmer, Howard G * ** * *
09300 Pardo, Luis Trabb G * ** *
09400 Pettit, Teri G * ** ** **
09500 Pfeil, Catherine N G * ** * * ** * **
09600 Phillip, James Robert P *
09700 Porter, Tom A * * ** * **
09800 Price, Bob G * * * * **
09900 Price, Gary P * * * * * *
10000 Pring, Edward G * * * ** *
10100 Rath, Marnie G * ** * *
10200 Rind, Didieu ** ** * * **
10300 Roberts, Janet G * ** * *
10400 Robinson, John G * ** * * * * **
10500 Rosenbloom, Kate A X
10600 Rosenbloom, Paul G * * * * ** **
10700 Samuel, Dan G * ** * * * *
10800 Samuelson, Ralph G *
10900 Scott, Carli G * ** * ** ** ** **
11000 Shattuck, William G * * * *
11100 Shenhar, Aanen ? * * *
11200 Terhune, Jim A *
11300 Thacker, Bill G * * * * * *
11400 Thompson, Clark G * ** ** * * * **
11500 vanMelle, Bill G * ** ** ** ** ** **
11600 Verducci, Joe P * ** *
11700 Vickery, David G * * * * **
11800 Wagner, Todd G * * * * *
11900 Wake, D. G * **
12000 Widdoes, Curt L. G * ** ** * ** * **
12100 Wilkins, David G * * * * * **
12200
12300 NAME G/P/A HW1 HW2 HW3 HW4 HW5 HW6 HW7 HW8
12400
12500
12600 Explanation of columns:
12700 NAME Last, First Middle
12800 G/P/A Grade or Pass/NC or Audit
12900 HW1 Homework assignment 1
13000 HW2 Homework Assignment 2
13100 .
13200 .
13300 .
13400 HW8 Homework Assignment 8
13500
13600
13700
13800 3. Homework Assignment #8
13900 Due Thursday, June 6, 1974. Homeworks turned in after June 4
14000 should be placed in the "L" mailbox on the second floor of Polya
14100 Hall, or in the LENAT mailbox at the AI Lab.
14200
14300 Do Problem A, and do one of Problems B and C.
14400
14500 A. Induction Programs
14600 Minsky and Papert describe Winston's network-growing program
14700 (CS224 Classnotes, MIT Progress Report) which learns concepts like
14800 ARCH and HOUSE.
14900 Formulate, in an English paragraph, what, e.g., a TENT would be
15000 in the Blocks World.
15100 Now draw the
15200 Winston-like network
15300 which represents this
15400 concept.
15500 Next, describe
15600 a training sequence of scenes which would cause Winston's program to
15700 develop this net. For each scene in the sequence, sketch the state
15800 of the TENT network just after exposure to the scene.
15900 Optional: Could a Winston-like net represent this concept: an
16000 EDIFICE is a TOWER, HOUSE, or ARCH. How Could Winston's program
16100 learn such a concept Does this indicate any difficult problems one
16200 must deal with when writing induction programs
16300
16400 B. DENDRAL
16500 Currently, the DENDRAL system is organized around a PLAN/GENERATE/
16600 TEST paradigm. How could it be organized around a GPS-like paradigm
16700 What would be the major advantages the major problems
16800
16900 C. DENDRAL
17000 You feed pictures like this one
17100 into DENDRAL. What knowledge must
17200 the program have in order to analyze
17300 them successfully That is, you
17400 are to describe and catalogue all the
17500 types of knowledge in DENDRAL.
17600 Optional: Draw a flow diagram of DENDRAL. Indicate what types
17700 of knowledge each part of the system has available. Why not let
17800 all parts of the program have access to all types of knowledge