17 09 12

Page 1

Soft Computing - September, 17 2012 • Use only the sheets provided to you, and in the two parts on separate sheets • Write clearly on the top left hand corner (like in the figure) surname, name, then, on the next line, enrollment number, date, then, on the next line, part of the exam (Part 1 or Part 2) and signature • Write CLEARLY by pen or pencil • Exam sheets not satisfying these simple rules, will result in exam penalization Cognome Nome Matricola Data Parte X Firma Esercizio n° Y feouf h rg oquyrbgrtpguh qreugh reugvr porihvd 0u8rh 3498uhg vrqe0uagh iuigfer iughpiufhvef fieuwh p3ruhfpew9ughvf piwe iuehf p iugh p iruwhg pr iughrw pghv ri3 rfbhv0r ewuhgf r3p3itugh p3ure ....... Esercizio n° Z feouf h rg oquyrbgrtpguh qreugh reugvr porihvd 0u8rh 3498uhg vrqe0uagh iuigfer iughpiufhvef fieuwh p3ruhfpew9ughvf piwe iuehf p iugh p iruwhg pr iughrw pghv ri3 rfbhv0r ewuhgf r3p3itugh p3ure ....... ... Part 1 ........................................................................................ 1.1. Fuzzy Systems [3/32]............................................ Describe at least two different types of membership functions that could be used to define fuzzy sets used as values for linguistic variables in the output of Mamdani fuzzy rules, and motivate their eventual selection for a fuzzy control system. 1.2. Reinforcement learning systems [7/32]........................................... Define a reinforcement learning system to make a robot learning to get out a maze. The robot can perceive obstacles with contact sensors positioned on the front, back, left, and right of the body, can move forward and backward, and turn on itself by 90 degrees. Please, define states, actions and reinforcement for this system, and select a reinforcement distribution algorithm. All choices have…