[Notes] Those who failed to get OK, do the following Select 30 data {(min)+(max)+(28 at random)}from each of the two attributes from banana database. Then create the table as follows: ------------------------------------------------------------ | | attribute-1 | atrrobute-2 | |------------------------------------------------------------| |membership | data | avg & std | data | avg & std | |------------------------------------------------------------| | | minimum | | minimum | | | | data-2 | | data-2 | | | Small | data-3 | avg = ( ) | data-3 | avg = ( ) | | | ... | std = ( ) | ... | std = ( ) | | | data-9 | | data-9 | | | | data-10 | | data-10 | | |------------------------------------------------------------| | | data-11 | | data-11 | | | | data-12 | | data-12 | | | Medium | data-13 | avg = ( ) | data-13 | avg = ( ) | | | ... | std = ( ) | ... | std = ( ) | | | data-19 | | data-19 | | | | data-20 | | data-20 | | |------------------------------------------------------------| | | data-21 | | data-21 | | | | data-22 | | data-22 | | | Large | data-23 | avg = ( ) | data-23 | avg = ( ) | | | ... | std = ( ) | ... | std = ( ) | | | data-29 | | data-29 | | | | maximum | | maximjum | | ------------------------------------------------------------ Send it to me by email till Sunday midnight (20 November)! ============================================================================================================= [A] AS-37 Sep Oct Nov total eval ------------------------------------------------------------------------------------------------------------- Saturday 11:30-13:30 03 10 17 24 01 Friday 15:10-16:40 (411) 09 23 30- 07 <-21 28 04 04 11 Tuesday 08:10-09:40 18 Saturday 16:50-18:20 (14) ------------------------------------------------------------------------------------------------------------- 2nd task on 04 November when you say R^n but how many was n? some who were given 3 seemed not repeated, i.e., n=0 ------------------------------------------------------------------------------------------------------------- Goishik Roman 4.5 4.5 5 5 4 5 3.5 1 C I x 3 3 4 42.5/40 OK Harhun Tsimafei 4.5 4.5 !5.1 5 4 5 3.5 3.5 a x 3 3 41.1 OK Grinyuk Dmitry 4.5 4.5 5 5 4 5 3.5 4 n a x 3 3 41.5 OK Gritsuk Ekaterina 4 4.4 4.5 5 4 5 3.5 4 c m 2.5 3 40.9 OK Kaminets Kiryl 4 4 5 5 4 5 3.5 4 e 3 4 41.5 OK Karabeika Kiryl 4 4 5 5 4 5 3.5 4 l i 3 4 41.5 OK Konovalov Vitaly 4.5 4.5 !4.5 5 4 5 3.5 4 e n 3 3 41.0 OK Lipotsev Anton 4.6 4.5 5 5 4 5 4.5 4 d 3 3 42.6 OK Lishko Aleksandr 4 4.4 4.5 5 4 5 3.5 4 ! a 3 3 42.4 OK Marchuk Viktoria 4 4.4 4.5 5.1 2 5 3.5 4 ! 3 4 41.5 OK Skovorodko Pavel 4 4 5 5 2 5 3.5 4 c 3 5 40.5 OK Siliuk Nikolai 4 4 4.5 5 4 5 3.5 4 o 2.5 3 41.5 OK Rapinchuk Igor 4.5 4.5 5 - 4 5 ^2.5 *5 n (2) 3 4 4.9 44.4 OK Navrosjuk Kostia 5 5 5 5 5 5 4.9 4.5 f 3 4 46.4 OK ------------------------------------------------------------------------------------------------------------- !=>10/29; ^=>11/04; *=>11/05 14 ------------------------------------------------------------------------------------------------------------- 09/03 create 2 metros both with constant speed, say 5, in one loop made up of 1000 pixels! 09/09 speed changes every step by adding -1, 0 or +1 at random. Show x, y, z, membership of speed of very slow, slow, medium, fast and very fast of both trains. 09/10 The same as yesterday, but membership function is not of speed but the rule {IF speed is medium AND distance is medium THEN break is very slow} 09/17 The same as AS-36-II on 09/16 but set of rules is {IF speed is small AND distance is medium THEN break is medium} {IF speed is small AND distance is small THEN break is strong} {IF speed is small AND distance is very small THEN break is very strong} and speed = 3, distance = 75, 100, 125, brake = 0,1,2,3,4,5,6,7,8,9,10 Show the table with 16 columns and 30 raws 09/23 The same as last time but 9 pairs of {speed = 4, 5, 6 and distance = 0, 500, 1000} Each pair should be added defizzified value of brake!! 09/24 The same as yesterday, 09/23, but a set of 9 rules of all combination of {speed = slow, medium, fast} and {distance = short, medium, long} brake should be determined according to situation, e.g. IF speed=fast AND distance=short THEN brake=very-strong Get de-fuzzified brake value for all combination of speed = 0, 10, 20 and distance = 0, 500, 1000 09/30 The same as 09/24, but a set of 25 rules of all combination of | {speed = veru slow, slow, medium, fast, very fast} and {distance = very short, short, medium, long, very long} 10/01 Get de-fuzzified brake value for all combination of speed = 0, 5, 10, 15, 20 and distance = 0, 250, 500, 750, 1000. Show 25 rules, table of speed-distance-brake of 25 combination, and 25 points in 3D coordinate. 10/07 11/04 Apply TS-formula and evaluate to Banana Dataset with 2/3 lines for designing and the rest for validation 2 Families of banana with each having 2 attributes. Membership functions should be ivery small, small, medium, large and very large for each of 2 attributes 11/04 Cluster Japanese 13 charachters ============================================================================================================= [C] AS-36-II Sep Oct Nov total eval ------------------------------------------------------------------------------------------------------------- Saturday 11:30-13:00 15 29 12 Monday 15:10-16:40 17(411) Thirsday 11:30-13:00 08 22 06 <----20 03==========> Friday 16:50-18:20 16 30 14 28 11 (310) ------------------------------------------------------------------------------------------------------------- Kurilenko Nikita 3 5 5 3 o 5 3 3.5 I 5 $4.5 4 5 48.0/48 OK Bogutskaya Yulia 3.5 5.2 5 5 5 5 5.1 3.5 5.1 4.5 4 5.1 56.0 OK Bryuzgin Alexander 4 5 5 @5 5.1 5 5.1 3.5 am 5 4 4 5.1 55.8 OK Burshtyn Alexander 4 5 5 5 (2) 5 5 3.5 4 4 4 5 51.5 OK Glebik Roma 4.5 5 5 @1 (2) 5 5 3.5 in 5 4.5 4 5 49.5 OK Zayats Igor 4 5.2 5 5 5 5 5 3.5 a 5 $4.5 4 5.1 56.3 OK Kolesnikov Dmitry 3.5 5.2 5 4.8 5 5 5 3.5 conf 5 4.5 ^3 5 54.5 OK Marchenko Anton 4 5.1 5 5 5 5 5 3.5 5 $4.4 4 5 56.0 OK Melnikov Artiom 3.5 5.1 5 @1 5 5 5 3.5 in 5 $4 4 5 51.1 OK Mironchuk Vlad 4 5.2 5 5 5 5 5 3.5 Poland 5 $4.5 4 5 56.2 OK ------------------------------------------------------------------------------------------------------------- @=>10/07; $=>10/27(minus advantage); ^=>11/04; *=>11/05 10 ------------------------------------------------------------------------------------------------------------- 09/08 same as the one on 09/03 but speed is not constant but starting with 10 and add -1,0,+1 at random every step 09/16 Create a table of speed = 11, distance = 550, break = 0,1,2,3,4,5,6,7,8,9,10 adding columns of mu(speed), mu(distance), mu(break), defuzzified value of these 10 break values, and mu(rule) Above should be done for each of 3 rules {IF x=medium AND y=small THEN z=strong} OR {IF x=medium AND y=medium THEN z=medium} OR {IF x=medium AND y=large THEN z=week}. Finally add the colume of mu(rule-1 or rule-2 or rule-3) by taking maximum value of mu(rule-1), mu(rule-2), mu(rule-3). 09/22 Same as task on 0916 basically but add defuzzified brake value for each of the 3 combinations of speed = 11, distance = 500, 550, 600. 09/30 The same as 09/22, but a set of 25 rules of all combination of {speed = veru slow, slow, medium, fast, very fast} and {distance = very short, short, medium, long, very long}. You should reasonably specify brake by yourself E.g. IF speed = very fast AND distance = very short THEN break = very strong. Get de-fuzzified brake value for all 25 combinations of speed = 0, 5, 10, 15, 20 and distance = 0, 250, 500, 750, 1000. Show 25 rules by a table, table of speed-distance-brake of 25 combinations, and 25 points in 3D coordinate. 10/06 The same as 09/30, but all 54 combinations of speed = {0, 2.5, 5, 7,5, 10, 12.5, 15, 17,5, 20} and distance = {0, 200, 400, 600, 800, 1000}. 10/14 Apply TS-Rule to iris-flower database with 3 Gaussian memberships of small medium large. With 7 rules in the form of {IF x_1=.. AND x_2=.. AND x_3=.. AND x_4=.. THEN y} (y=A, B, C, else) validate this rules with data for validation 10/15 Obtain 5 Gaussian membership function of very small, small, medium, large, very large from the iris flower data. design 3 rules for each family and apply TS-formula. 10/17 Apply TS-formula amd evaluate to Glass Dataset with 50 lines for designing and the rest for validation 7 Families of glass with each having 10 attributes. Membership functions should be small, medium and large for each of 13 attributes 10/28 Correction of task on 10/17 by adding (avg) and (std) for each membership function. Also show how these (avg)s and (std)s are ditermined from the data Alao cluster 10 Russian characters you like. 10/29 Cluster 26 Roman alphabets 11/11 Design 5 membership function of each of 2 attributes of big banana database 11/12 Time-series prediction by TS-Rule with 5 membership ============================================================================================================= [B] AS-36-I Sep Oct Nov total/minimum Eval ------------------------------------------------------------------------------------------------------------- Saturday 11:30-13:30 08 22 05 Thirsday 11:30-13:00 15 29 13 27 Friday 16:50-18:20 09 23 07 <-----21 04 (11) Monday 13:30-15:00 (411) 17 Monday 16:50-18:20 additional ------------------------------------------------------------------------------------------------------------- The task on 11/04 should be included R^0 and n such that R^n = R^(n+1) ------------------------------------------------------------------------------------------------------------- Muzyka Aleksandr 5 5.1 5 3.5 5.1 4 5 3.5 I 5 3 5.1 49.3/44 OK Nikolaenko Anastasiya 5 4.9 5 3.5 5 4 5 4 4.5 3 5 48.9 OK Fedorov Danil (2) 2.5 5 3.5 5 4 5 3.5 am 5 3 5 4 47.5 OK Palik Mark 4.5 2.5 5 3.5 5 4 4 3.5 4.5 3 5 44.5 OK Potapiuk Aleksei 5 4 5 3 5 4 (2) 3.5 in 5 3 5 44.5 OK Aksiutchyk Vitali 4 4 5 @1 (2) 4 4 3.5 a 3.9 3 5 4 5 48.4 OK Timohin Valentin 4 4 5 3.5 3 4 4.9 3.5 conf 4 3.5 5.1 44.5 OK Chekun Roman 5 4 5 5 5 4.1 5 4 5 3 5.1 50.2 OK Shurpo Dmitry 5 5 5 3.5 2 4 4.5 3.5 in 4 3 5 44.5 OK Trotsiuk Aleksei 4.9 5 5 3.5 5.2 4 5.1 4 Poland 5 5.1 5.1 51.9 OK Savitsky Anton $5 5 - (2) 5 4 4.5 3.5 2 5 2 5 43.0 ok Savchuk Artem 5 4.5 5 (2) 5.1 5.1 5 4 5.1 4 5.1 49.9 OK ------------------------------------------------------------------------------------------------------------- @=>09/30; $=>10/28 ------------------------------------------------------------------------------------------------------------- 09/09 Create 2 metros both in one loop made up of 1000 pixels. Starting with speed=10, it changes every step by adding -1, 0 or +1 at random. Show x(speed), y(distance to the front car), z(break), membership of speed of very slow, slow, medium, fast and very fast of both trains. 09/15 The same as before but show the membership value of 3 rules {IF x=medium AND y=small THEN z=strong} OR {IF x=medium AND y=medium THEN z=medium} OR {IF x=medium AND y=large THEN z=week} instead of membership value of speed Brake z is not 0 like last time but => {if y>=14 then z=1, x=x-1} & {if x<=50 then z=9, x=x-9} Every step, Speed x is added +2, 0, -2 at random, not +1,0,-1 like last time 09/23 The same as the task on 09/22 for AS-36-II but a set of 3 rules is {IF x=very small AND y= very small THEN z=very strong} OR {IF x=small AND y=small THEN z=strong} OR {IF x=medium AND y=medium THEN z=medium} for 9 pairs of {speed = 0, 1, 2 and distance = 0, 100, 200} 09/29 The same as the task on 09/23, but under a set of 9 rules of all combination of {speed = veru slow, slow, medium, fast, very fast} and {distance = very short, short, medium, long, very long}. 10/07 The table should be completed with more points e.g. speed = {0,1,2,3,...,20} and distance = {0,50,100,150, ..., 1000}. Then two metoro should be controlled by this table. Show 5 typical snapshots from 2 metoro animation. the value of speed, distance, brake of both trains should be included! 10/08 Iris Data => divide 17x4x3=204 data into 3 groups of small medium large and calculate avg and std for each of 3 groups. Then draw 3 Gaussian membership function of small, medium, and large in x-y coordinate Create 3 rules for each family A, B, C by your feeling like IF x_1 = small AND x_2 = large AND x_3 = medium THEN A. Using 2nd 8 data to evaluate your 3 rules good or not by check OK NO False alarm. E.g. -------------------------------------------------------------------------------- Family A Family B Family C -------------------------------------------------------------------------------- Rule-1 Rule-2 Rule-3 -------------------------------------------------------------------------------- No. A or B or C or other A or B or C or other A or B or C or other -------------------------------------------------------------------------------- #1 A B C good #2 A A other not good #3 other B other not good #4 B A C not good #5 A B C good .................. #8 A B C good --------------------------------------------------------------------------------- success rate 50% 40% 90% 62% --------------------------------------------------------------------------------- 10/13 Apply TS-formula and evaluate this 3 rules 10/17 Apply TS-formula amd evaluate to Wine Dataset with 50 lines for designing and the rest for validation 3 Families of wine with each having 13 attributes. Membership functions should be small, medium and large for each of 13 attributes 10/27 Again classify Wine, but 5 membership very small, small, medium, large, and very large, with 7 rules! 11/04 Cluster Russian 36 alphabets! 11/05 Prediction of x(t+1) from Time-Series dataset x(t) to x(t-60) (1) create 3 membership function of small, medium, large from dataset (give me 3 graphs and avg and std each) (2) design 9 rules such as IF x(t-1) = large AND x(t) = medium then y_i (3) create a table such as ------------------------------------------------ t x(t) x(t-1) how-big x(t-2) how-big y ------------------------------------------------ 60 ? 346 large 250 small 59 346 250 small 200 small 58 250 200 small 300 medium 57 200 300 medium 400 large ............ 02 120 100 small 600 large 01 100 600 large --- --- 00 600 ------------------------------------------------- (4) Apply TS-rule (See p.34) and fill the column y above (5) predict x(t) by y such as if y>5 then large, if 3 without checking theri result. If you don't like this, attend another group on another day. Then the point will be replaced with a new. All those who don't have OK yet, you should attend practice on 14 November from 16:50 to 18:20 at 310 This will be your last chance!